MAHANGU POST-HARVEST SYSTEMS

A SUMMARY OF CURRENT KNOWLEDGE ABOUT PEARL MILLET POST -

HARVE ST ISSUES IN NAMIBIA

Research Report

prepared for Ministry of Agriculture, Water and Rural Development

Directorate of Planning and

Nam ibian Agronom ic Board

by Michel Malle t & Pierre du Plessis

CRIAA SA-DC (Namibia)

March 2001

Mahangu Post-Har vest Syst ems A Summary of Current Knowledge about Pearl Millet Post-Harvest Issues in Namibia Research Report prepared for Ministry of Agricul ture, Water and Rural Development Directorate of Planning Private Bag 13184, Windhoek Tel. (061) 208 7111 Fax (061) 208 7767 as part of the project Technical Support to the Small Scale Mahangu and Other Grain Processing and Marketing Industry in the Northern Communal Areas of Namibia Project managed by the Namibian Agronomic Board P.O. Box 5096 Ausspannplatz, Windhoek Tel. (061) 22 4741/2/3 Fax (061) 22 5371 Report written and edited by Michel Malle t and Pierre du Plessis CRIAA SA-DC (Namibia) P.O. Box 23778, Windhoek Tel. (061) 220117 Fax (061) 232293 Email: cri aawhk@iafri ca. com.na <mailto: cri aawhk@iafrica. com.na>

CONTENTS ACKNOWLEDGEMENT S ............................................................................................................... v ACRONYM S AND ABBRE VI AT IONS ......................................................................................... vi ABST RACT ..................................................................................................................................... vii CHAPTER 1 INT RODUCT ION........................................................................................... 1 Background.......................................................................................................................................... 1 Pearl Millet in the World Today - A Brief Overview ........................................................................ 2 Grain Post-Harvest Systems - A Definit ion ....................................................................................... 4 CHAPTER 2NAT IONAL MAHANGU PRODUCT ION AND NATIONAL CEREAL CONSUMPT ION ............................................................................................ 5 National Mahangu Grain Production .................................................................................................. 5 National Mahangu and Cereal Consumption ..................................................................................... 7 CHAPTER 3ON-FARM MAHANGU POST -HARVE ST SYST EMS ............................. 10 Harvesting and Drying ..................................................................................................................... 10 Next Season’s Plant ing Seed ............................................................................................................ 11 Threshing and Winnowing ............................................................................................................... 12 Pre-storage threshing ....................................................................................................................... 13 The different methods of threshing used in the Northern Communal Areas ...................................... 13 Economic overview of threshing ...................................................................................................... 14 On-Farm Storage .............................................................................................................................. 16 Household Food Processing ............................................................................................................. 18 Mahangu primary processing ........................................................................................................... 18 Mahangu secondary processing and food preparation

....................................................................22 Household Grain Market ing ............................................................................................................. 23 Concluding Remarks ........................................................................................................................ 24 CHAPTER 4GRAIN TRADING AND PROCE SSING SE CTOR ................................... 26 Mahangu Grain T rade and Markets .................................................................................................. 26 Local barter or exchange market ..................................................................................................... 26 Informal trade market ....................................................................................................................... 26 Formal trade ..................................................................................................................................... 26 Inter-regional and international trade ............................................................................................. 27 E = mc² (Empowerment = mahangu x cash²) ? ............................................................................... 28 The Service Milling Sector ............................................................................................................... 29 Distribution of service m ills and evolution in the past years ........................................................... 29 Technology used and technical constraints ...................................................................................... 30 Financial viability of service milling ................................................................................................ 32 Economic importance of service m illing .......................................................................................... 32 The Commercial Milling Sector ....................................................................................................... 33 Regional distribution of small commercial mahangu mills and installed capacity ......................... 33 The different commercial methods of processing mahangu ............................................................ 34 Technology used and technical constraints on commercial mahangu processing .......................... 34 Financial viability of commercial mahangu milling ........................................................................ 35 Further research and development needed ..................................................................................... 35 Utilisation of commercial mahangu mill products ........................................................................... 36

CHAPTER 5 MAHANGU PROMOTION POLICY HIGHLIGHT S AND DISCUSSION 38 Inst itut ional arrangements, progress review and re-planning ........................................................... 38 Primary agricultural production ....................................................................................................... 38 Post-harvest processing and grain storage ........................................................................................ 39 Mahangu grain marketing and processing ........................................................................................ 39 Value-addin g and product development ........................................................................................... 40 Conclusion .........................................................................................................................................40 ANNEX 1 BIBLIOGRAPHY ............................................................................................................ 41 ANNEX 2 Maps of milling services in Caprivi, NCRs, Kavango ............................................ 48 - 50 TABLES Table 1 Namibia’s Mahangu and Sorghum Grain Production Estimates in metric tonnes (t): Production years 1990/19991 - 1999/2000 (all ran-fed production).................................................6 Table 2 Namibia’s Annual Cereal Consumption as Food (1990 - 2000) .......................................... 7 Table 3 Namibia’s Cereal Production, Imports and Utilisation in metric tonnes: Marketing years 1991/92 - 2000/01 .............................................................................................................................. 7 Table 4 Farmer’s ranking of grain traits compared to the local variety (1993) ............................. 12 Table 5 Average composition of pearl m illet grains (percentage of total weight) ........................... 18 Table 6 Physical characteristis of Okashana-1 and local mahangu varieties (Kavango) ............... 19 Table 7 Comparison of Okashana-1 with local variety by hand-pounding (on six runs) ................ 20

Table 8 Traditional Preparationof Mahangu Flour ......................................................................... 21 Table 9 Proportion of households buying grain, Dec. 19992 .......................................................... 23 Table 10 Proportion of grain purchased, Dec. 1992 ....................................................................... 23 Table 11 R&D needs for addressing farmers’ post-harvest constraints .......................................... 25 Table 12 Comparative prices (N$/tonne) of Maize Meal in North Central Namibia (1999) ........... 33 Table 13 Average composition of cereal grains (percentage of total dry weight) ........................... 34 Table 14 Analysis of a mahangu flour sample from Kamalanga Mill (May 1998) .......................... 36

ACKNOWLEDGEMENTS The authors wish to thank the following persons for their valuable contributions in reviewing the draft report : Ben Bennett of the Directorate of Planning in MAWRD, Jür gen Hoffmann of the Namibian Agronomic Board and Arne Larsen of the Research Extension Management Programme (REMP) within MAWRD. We also wish to thank all those who contributed over the years to our bet ter knowledge of mahangu post-harvest in Northern Namibia, who cannot all be named here. A special note of appreciat ion to Frederika Amutse of CRIAA SA-DC for her Oshindonga and Oshikwanyama language support , as well as for sharing her extensive knowledge of tradit ional food processing. It must be noted, however, that this research was based on available literature (in a printed form) and on the professional experience of the authors. As the research did not include specific fieldwork in the North, unpublished materials, ‘grey literature’ and farmers’ knowledge could consequently not be systematically collected. (Maybe this could be considered in future...)

ACRONYMS AND ABBREVIATIONS CIRAD Centre de coopérat ion internationale en recherche agronomique pour le développement (France) International Co-operation Centre in Agronomic Research for Development CRIAA SA-DC Centre for R esearch-In formation-Action in Africa, Southern Africa - Development and Consulting CSIR Council fo r Scienti fi c and Indust rial Res earch (South Africa) DCD Division o f Co-operative Development (MAWR D) DfID Departm ent fo r International Developm ent (UK) DoP Directo rat e of Planning (MAWR D) FAO Food and Agricultu re Organis ation (UN) FSRE Farming Systems Research and Extension ha hectare (10 000 m²) hp hors e-power (0.7355 kW) ICRISAT International Crops Research Institute fo r Semi-Arid Tropics kJ kiloJoule kW kiloWatt MAWRD Ministry o f Agriculture, Water and Rural Development MMIU Mahangu M arketing Int elligence Unit MRC Multidisciplinary Res earch Cent re (UNAM) NAB Namibi an Agronomi c Board NCAs Northern Communal Areas NCRs North Central Regions (Oshana, Oshikoto, Ohangwena and Omusati ) NDC Namibi a Development Corporation N$ Namibi a Doll ars NEWFIU Namibi a Early Warning and Food In fo rm ation Unit (MAWR D) NNFU

Namibi a National Farmers’ Union NRC Namibi a Resource Consult ants NRI Natu ral Res ources Institute (UK) Orstom Institut de recherche pour le développem ent - IRD (fo rmerly Orstom) (France) PTO Power take-o ff RDC Rural Development Centre SACU Southern Africa Customs Union SADC Southern African Devel opment Community t tonne (1 000 kg) UNAM University o f Namibi a UNIC EF United Nations Children Fund

ABSTRACT In order to share information and disseminate findings, this paper condenses all the reports produced to date under the project « Technical Support to the Small Scale Mahangu and Other Grain Processing and Marketing Industry in the Northern Communal Areas of Namibia ». A list of reports appears in Annex 1. Results of other relevant work on pearl millet post-harvest issues are also included (see Bibliography). The present world situat ion around mahangu (pearl millet) production and trade is described an d a working definit ion of post-harvest systems is presented. National cereal production, including mahangu, is compared with national cereal consumption. T he role of imports is quantified and described. Despite a strong consumer preference for mahangu in some sectors, cheap maize meal is very important because the price of cereals plays a crucial role in the food security, nutritional status and living standard of especially the poorest households. Exist ing on-farm post-harvest systems are described for harvest ing, drying, seed select ion, threshing and winnowing, on-farm storage, household food processing and grain market ing, and related research and development needs are identified. T he current trade in mahangu grain in various markets is characterised and part ially quantified. T he service and commercial mahangu milling sectors are described and quantified, and the constraints they face are elucidated (see Maps 1-3). Finally the summarised information is discussed in relat ions to mahangu promotion policy. It is concluded that better post-harvest pract ices and technology can increase the effect ive nat ional harvest and the efficiency of processing, but that sustainable growth of the secondary and tert iary mahangu sectors will ult imately need to be supported by increased primary production and/or more reliable and cheaper imports of mahangu grain, especially in drought years. However, many of the constraints on mahangu post-harvest systems have been identified and addressing these constraints is there fore a priority area for immediate intervention. CHAPTER 1 INT RODUCT ION Background The Ministry of Agriculture, Water and Rural Development (MAWRD), through its Directorate of Planning (DoP), has embarked on a project « Technical Support to the Small Scale Mahangu and Other Grain Processing and Marketing Industry in the

Northern Communal Areas (NCAs) of Namibia ». T his project resulted from recent work, which recommended that Government focuses its support on the promotion of an emerging grain processing sector in the northern crop growing communal areas. T he project , which started in June 1998, is managed by the Namibian Agronomic Board (NAB) and supported by a Steering Committee. Through a tendering process, CRIAA SA-DC (a not-for-gain Namibian-registered development and consult ing organisation) was commissioned to implement this project . Phase I was a comprehensive baseline study, which identified a large n umber of mills in Northern Namibia, as well a s technical and operat ional difficult ies. As a result of the findings the Steering Committee commissioned a second phase (1999-2000) to invest igate the technical constraints and economics of small-scale mahangu milling. As part of the terms of reference of Phase II, the consultants were required to provide costed proposals for research into alleviat ing some of the crit ical technical constraints that hold back the development of farm and village level mahangu processing in NCAs. T he outcome was the following recommendations: A range of adaptive technical research addressin g key constraints to mahangu processing by small farmers and small millers; Preparation and delivery of t raining to small millers to improve the efficiency of and financial returns from milling; A number of on-farm post-harvest pract ices identified durin g the study could be significantly improved to increase the volume of mahangu available for sale as surplus. The list of project reports to date is shown in Annex 1. The present research paper aims at condensing all the project reports to date in order to share information and disseminate findings and recommendations on the topic. Results of other relevant research and development work conducted in Namibia, or elsewhere, on pearl millet post-harvest issues are also included in this document, as far as they are known to the authors (see Bibliography in Annex 1) and relevant to the purpose of this synthesis. Government policy and programmes for commercialising mahangu cannot be ignored in the context of this paper, but will not be its main focus. T he National Agricultural Policy (1995) emphasises that the commercialisat ion of pearl millet is a strategy to alleviate poverty and improve the standard of living of small-scale farmers in the Northern Namibia. T he MAWRD has also a clear vision that improved ut ilisat ion of this valuable resource can benefit household food security, agricultural industries and small and micro-enterprise development. It can also lead to novel agronomic and food industries development, since mahangu is tolerant to the marginal agricultural condit ions prevailing in the North of Namibia, and can offer significant opportunities for strengthening the whole agricultural sector in Namibia. A workshop recently held at Mokuti Lodge stressed these points. Pearl Mille t in the World Today - A Brie f Overview Pearl millet, commonly named Mahangu in Namibia, is one of the most important human staple cereals grown in the world today after wheat , rice, maize, barley and sorghum. Its taxonomic name is Pennisetum glaucum (L.) R. Br. but it is also known as P.

americanum and P. typhoides. Bulrush millet , cattail millet and candle millet are also used as synonyms for pearl millet . However, pearl millet should not be confused with the other millets, such as finger millet (Eleusine coracana), foxtail millet (Setaria italica), common or proso millet (Panicum miliaceum), and fonio (Digitaria exilis) of West Africa, all of which are not grown in Namibia. Pearl millet is a t radit ional staple food crop of the semi-arid tropical zones of Africa and Asia, and it is cult ivated in areas with a rainfall ranging from 150 mm to 800 mm per annum (CIRAD/Orstom, 1997). It has a nutritional value superior to other cereals like maize, wheat or rice: it is digested more slo wly and thus delays hun ger pangs, it has a higher content of proteins and lipids, and its amino acid balance is bet ter (Labetoulle, 2000). Pearl millet is the most drought tolerant cereal and can grow in soils with low fert ility, which are unsuitable to other cereals. Being part icularly resistant to moisture stress and adapted to high temperatures, it has a low yield, which also explains its high nutrit ional value compared to other cereals. In t raditional African agricultural systems, pearl millet is very often intercropped with pulse plants, such as groun dnuts or cowpeas, and with cucurbits. Typically, livestock plays a crucial part in pearl millet based traditional farming systems. Pearl millet stover is a valuable source of animal feed, making it a dual-purpose crop for subsistence farmers (Blench, 1997). In some regions, the stover is also used as roofing or fencing materials. Pearl millet furthermore has excellent storage qualit ies, giving it an advantage over other crops where inter-annual food security is necessary (Tyler & Bennett, 1993). In Africa, pearl millet is typically a “ low-input/low-output” rain-fed crop of marginal agricultural areas. Average yields are est imated around 0.6 tonnes per hectare (around 0.8 t/ha in India) but range between 0.3 t /ha in Southern Africa to over 0.8 t /ha in Nigeria and Uganda under wetter condit ions (NRC, 1997). It must be noted that a t radit ional pearl millet variety specific to the Senegal River Valley is gro wn counter-seasonally in wet soils left behind by recedin g floods (CIRAD/Orstom, 1997). This is also the case with some sorghum variet ies in West Africa, as well as in Kavango along the river and East Caprivi in the Zambezi flood plains (where maize is also grown this way). Although accurate stat ist ical figures on pearl millet production are not available, FAO est imated that it was gro wn on over 30 million hectares in 1994, mainly in developing countries, and that world production was aroun d 28 million tonnes. Africa accounted for 20 million hectares and a production of 11.9 million tonnes, and India, the leading world producer, 13.7 million hectares and 11 million tonnes. T he world t rade of pearl millet is only around 0.25 million tonnes a year and less than 1% of the world’s millet production is internat ionally traded. In Africa, the largest producers of pearl millet are situated in the Sahel in West-Central Africa: Nigeria, Niger, Mali, Senegal, Burkina Faso and Chad. In East Africa, Uganda and Sudan are the largest producers. Only 5% to 10% of African production is est imated to enter the commercial market (FAO/ICRISAT, 1996), while in India, where pearl millet is grown commercially, an est imated 15% to 20% is marketed. In the dryer parts of Southern Africa (in Angola, Namibia, Zimbabwe, Botswana, Zambia and South Africa), millet is t radit ionally grown as a staple and surpluses are hardly traded. However, cross-border t rade from Southern Angola to Northern Namibia has been reported as significant in some years, even in the absence of official t rade and customs

figures. In Zimbabwe, pearl millet cult ivat ion has been declining from over 250 000 ha in the 1980s to less than 150 000 ha in recent years, with average yields also dec lining. « With 51 000 t produced annually and a yield of 0.4 t /ha (averages over the past 5 years), pearl millet accounts for approximately 2.4% of coarse grain production in Zimbabwe, where it is exclusively grown on smallholder and reset t lement farms. In the 1998/99 marketing year, formal market deliveries were only 310 t (for a production of 29 900 t that year), 300 t to a commercial opaque beer brewer and an est imated 10 t as poultry feed. The Grain Market ing Board, a parastatal mandated to act as a residual buyer, has purchased virtually no pearl millet during the past 2 market ing seasons, and is unlikely to purchase any again this year (2000). ICRISAT ’s surveys generally indicate that pearl millet prices are 10 to 20% higher than the price of maize, even in areas of the country where pearl millet is a main staple. Pearl millet could potent ially be used significantly more in the food and feed industries of Zimbabwe, but given its higher grain price, it is best targeted as a speciality product in urban markets and as a staple in rural producing areas » (adapted from Rohrbach, 2000). In a study commissioned by MAWRD/NAB in 1997, The World Market for Mille t and the Potential for Imports into Namibia, to determine the possibil ity of import ing mahangu to facilitate the establishment of a substant ive processing industry, NRC consultants concluded as follows. « Most of the millet t raded internat ionally is proso millet , which is used mainly as birdfeed. Virtually all of the world’s production of pearl millet is by subsistence farmers for own-consumption; it is rarely t raded. Countries bordering Namibia are not regarded as suitable for securing millet when Namibia is short . The countries experience a similar climate and more than likely would also be in deficit . Also, most of the production in neighbouring countries is for subsistence and not for commercial market ing. West Africa is a significant producer of pearl millet . However, it is all grown well in-land in areas bordering the Sahara desert , about 1 000 to 1 500 miles from the sea on very poor roads. The high cost of road transport to a seaport , as well as sea freight and import dut ies would make West African pearl millet prohibit ively expensive. (...) The only country that could supply Namibia with pearl millet in most years is India. Price models indicate that Indian pearl millet could be landed in Oshakati for prices between US$ 277 to 335 per tonne (N$ 1 330/t to N$ 1 608/t , 1997 prices). However, even at these factory gate prices, millet flour could be retailed at N$ 3.10/kg to N$ 3.40/kg, which is over twice as much as maize meal. (...) Therefore , i t must be concluded that i t is impossible to import pearl mille t and process i t into flour in Namibia at a price which is competitive with maize meal . » In the world, most pearl millet is grown for grain used for human consumption, but there is increasing interest in ut ilising it as a forage crop and feedgrain crop. This trend is developing in drier parts of the USA, Australia, South Africa and South America. In Brazil, pearl millet has been rapidly adopted by farmers as the ideal cover crop-cum-mulch for no-t ill soybean production. Over 1 million hectares are reported to be sown in Brazil, especially on acid savannah soil (SAT News).

Pearl millet is often considered as the subsistence crop of last resort in agricultural marginal areas and host ile environments, the production of which is virtually unimprovable. These considerat ions are only part ly true. First ly, smallholder farmers can potent ially increase productivity, or render production more reliable, with genet ically improved varieties (ICRISAT has been playing a major role for t ropical areas) and improvements in cult ivat ion techniques, such as draft animal power, as exemplified by the Indian experience. Pearl millet shows a high response to limited fert ilisat ion with manure and/or nitrogen and phosphorus (Matanyaire, 1998), which are most often deficient in t ropical soils. However, increasing the yield per hectare is taken as the ult imate criteria by most extensionists for improving smallholder cropping systems, while farm labour is most often the major bott leneck for peasant farmers, when farm land is not dramatically scarce and farming techniques are not much mechanised. T he productivity of farm work (quantity harvested per unit of work applied) is often a more relevant criterion from a producer’s perspect ive, but more difficult to measure (by researchers). Secondly, a large number of African countries, especially in West Africa, have surprisingly managed to increase per capita production of pearl millet (as well as sorghum and other grains) since the last great drought of 1982-1985. T he liberalisation of cereal markets joint ly init iated in 1986 by Sahelian countries, with accompanying measures and substant ial support from donor countries, had a posit ive impact on private grain t rading operators and spin-off effects on primary agricultural production by rural producers. T he devaluat ion of the CFA franc on 14 January 1994 unleashed the development of emerging small-scale processin g enterprises providing new local products for the urban consumer markets (T. Dauplais, pers. comm .). In Mali, where cereal production increased by over 50% after liberalisat ion in the late 1980s, the restructured market for locally produced grains, including pearl millet , has been managed by a large network of private traders (the State providing market information and keeping strategic reserve stocks). Village a ssociat ions, mobile ‘collectors’ in rural areas and markets, ‘assemblers’ operat ing at urban or large rural market level, wholesalers and semi- wholesalers an d retailers in urban areas, as well as processors, all played an important role in transferring cereals from main production areas to consuming areas and redistribut ing cash income to rural producers (Kanté, 2000). However, liberalisat ion did not render pearl millet price-competit ive with maize or rice. The object ive of support ing the prices of local cereals was abandoned in the late 1980s mainly because of the difficulty of financing the stock levels necessary to maintain prices in surplus years (Coulter, 1994). Grain Post-Harvest Systems - A Definition Post-harvest of cult ivated cereals embraces all operations from harvest to final ut ilisat ion of the grain (food preparat ion being the main ult imate use for the purpose of this paper), including harvest ing, drying, threshing, cleaning, storing, grading, market ing and processing. From a technological angle, post-harvest operat ions are divided into crop processing (or primary processing) and food processing (or secondary processing). However, a technology analysis is not sufficient to characterise post-harvest systems, even if the important subject of farmers’ knowledge sy stems is included. Economic issues are also

crucial (demand, prices, markets, etc.). A micro-economic angle will focus on post-production at the individual farmer’s level, while a meso- or a macro-economic angle will look at processing and market ing at regional or nat ional level. Post-harvest issues are therefore complex, involving different inter-linked levels of understanding, and requirin g mult i-disciplinary and mult i-level analyses. Hence the reference to a systemic approach in « post-harvest systems ». CHAPTER 2 NAT IONAL MAHANGU PRODUCTION AND NAT IONAL CEREAL CONSUMPT ION National Mahangu Grain Production Mahangu is the main crop of the Northern Communal Areas (NCAs), specifically of the North Central Regions (NCRs), Kavango and the western part of Caprivi, where it is grown un der rain-fed condit ions. T o a limited extent , it is also grown in some parts of the Otjozondjupa region, especially in the former West Bushmanland, and it has recently been cult ivated by some ‘commercial farmers’ mainly in the T sumeb-Grootfontein-Otavi Triangle. In the NCAs, mahangu and sorgh um are very often grown in the same fields, not necessarily inter-cropped (sorghum being grown in some patches of the main mahangu fields). This makes it difficult (and part ly irrelevant) to segregate mahangu and sorghum hectarages. On average, however, sorghum is est imated to represent around 10% of the combined mahangu-sorgh um production. Sorghum yields are est imated to be similar to those of mahangu. In the NCAs there are around 120 000 mahangu/sorghum growing households cult ivat ing these crops on between 250 000 and 280 000 hectares (NEWFIU), which gives an average cult ivated area of between 2 and 2.5 ha per household. However, this average hides an uneven distribut ion between large cult ivators (say over 5 ha cropped) and very small (less than 2 ha). A limited number of communal farmers, mainly in the Mangett i area in Oshikoto and around Katjinakat ji in Kavango, cult ivate much larger mahangu fields and are regular surplus producers. Table 1 overleaf gives the ‘official estimates’ of mahangu and sorghum grain production for the past ten years (as published by NEWFIU). T he ten-year records (production years from 1990/91 to 1999/2000) give an average production of 58 300 tonnes per annum and an average yield of 230 kg per ha. T he past five-year averages give an annual production of 70 900 tonnes and a yield of 250 kg per ha. However, annual production has fluctuated significantly according to the weather condit ions from a low 17 200 tonnes in 1991/92 to a peak of 117 100 tonnes in 1996/97. According to these data, annual mahangu/sorghum grain production per capita in northern Namibia has varied between 20 kg and 125 kg in the past ten years. However, these official est imates of mahangu production need to be considered with care. « T hough efforts have been made to improve the quality of these estimates, the regional aggregates may underest imate the actual levels of sown area and yield. Farm surveys

indicate that the nat ional est imates published by NEWFIU underest imate the pearl millet area per household. For example, according to Matanyaire (1996), the average area sown to pearl millet in northern Namibia is 3.5 ha per household. The mean for the NCRs was 3.1 ha compared with 4.1 ha for Kavango. However, NEWFIU (1997) est imated the nat ional average to be only 2.1 ha per household. » (Rohrbach & al., 1999)

As it is generally considered that for a cereal-based diet a consumption of 120 kg of grain per annum per capita is required to cover basic nutritional requirements, it is concluded that the NCAs have on average only reached global cereal self-sufficiency in one year durin g the last decade (MA WRD, 2000). National Mahangu and Cereal Consumption Namibia’s annual cereal consumption has averaged about 204 000 tonnes for food use in the past ten years (and over 230 000 tonnes in the last five years, due to populat ion growth). Maize represents around half of the total cereal consumption as foodstuff, in the form of white maize. Wheat consumption, which current ly accounts for 20% of cereal consumption in Namibia, is believed to be growin g rapidly as the country’s populat ion becomes more urbanised (Fowler, 1997). Mahangu and sorghum represent on average 30% of nat ional cereal food utilisat ion. Mahangu is far from meeting the total cereal consumption requirements of the country. It does not even meet the needs of northern Namibia where over half of the populat ion lives. Consumption is met by both domestic production and imports. In most years Namibia has to import a considerable port ion of its cereal requirements for food use, on average 111 000 tonnes per year in the past ten years, and 124 000 tonnes per year in the last 5 years (see Table 2 below and Table 3 overleaf). South Africa is the main producer country from which white maize and wheat have been imported. Table 2 Namibia’s Annual Cereal Consumption as Food (1990 - 2000) Average Total cereals Mai ze Mahangu/sorghum Wheat 10 years 204 110 t 100 % 100 660 t 49 % 58 190 t 29 % 45 260 t 22 % Last 5 years 230 660 t 100 % 111 680 t 48 % 70 620 t 31 % 48 360 t 21 %

Source: Namibian Early Warning & Food Information Unit (MA WRD) The nat ional average per capita consumption of grain is est imated at around 120 kg - 135 kg per annum. But it is likely that this figure is higher for the rural populat ion in the NCAs ( where the diet is cereal-based) than for the rest of the country (where the diet is more diverse). Mahangu is the preferred cereal for a large part of the NCRs and Kavango populat ion (as well as for most people originat ing from these regions and residin g/working in southern parts of the country). A recent consumption survey (Kandando & Ngwira, 1999) on a small sample of family households in three urban centres of the NCRs and in Windhoek confirmed that more than 95% of respondents prefer mahangu to all other grain-based products. T he survey results also showed that over 90% of the respondents consume mahangu in the form of st iff porridge and Oshikundu (non-alcoholic fermented drink), but that a large proport ion of respondents also eat rice (80%), bread (78%), breakfast cereals (57%), biscuits (50%) and pasta (43%), while 70% also drink Omalovu (sorghum beer). The survey also revealed that 76% of the respondents consumed their own mahangu (but a majority produced le ss than 500 kg per year), and that the majority of respondents (69%) were spending between N$ 20 and N$ 50 per month on buying mahangu.

There has been no comprehensive food consumption survey and market research in Namibia, especially related to mahangu and other cereal products, apart from the UNAM survey mentioned above (which had a limited coverage), and consumer preference tests carried out for the launching of a new industrial mahangu flour product . T his issue has been highlighted recently by a team of CIRAD consultants on a mission to Namibia (Goli & Ndiaye, 2000): « It is a crucial issue to have a bet ter knowledge of food consumption habits and trends in Namibia, especially in town, and identify the actual market (size, segments and constraints) and demand (typology of consumers, quality criteria, food preferences, price acceptability, brand image and reputation) for the different processed products. » Households in the NCAs balance their cereal production deficit by purchasing maize meal in most years (and every year for a good proport ion of them). As Keyler (1996) reported from his study durin g 1992/93 in former Ovambo and Kavango regions: « Research findings demonstrate that few farmers are millet self-sufficient and only 10% of millet production is commercially traded. Most people in the study zone prefer millet over maize. But millet’s scarcity and high price leave imported maize to dominate the commercial food market and att racts millet imports from Angola. » Commercial maize demand in Northern Namibia depends largely on the level of the mahangu harvest of the year (and also in Caprivi on the size of the local maize harvest). Maize demand also varies durin g the course of the year, with peak demand usually start ing in December/January and stretching to the mahangu harvest around May. In this regard, maize has always had advantages over mahangu: it has been available in a processed form (mealie meal) throughout the year in urban and rural areas, well distributed through a competitive network of wholesalers, supermarkets and small shops, and it has been sold at stable and att ract ive prices compared to locally available mahangu flour or even grain. In 1997, researchers est imated from commercial processing sources that the annual maize meal consumption in the NCRs over the past few years was between 15 000 and 45 000 tonnes, depending on the year (Mallet & El Obeid, 1997). The nat ional annual per capita maize meal consumption is around 65-70 kg (NEWFIU). In the Northern Regions expenditure on maize meal as a proport ion of total expenditure is two and half t imes that of the whole of Namibia. T he poorest half of households in northern Namibia spent around 40% of their household budget on maize meal (Low, 1995). Hence the importance of both imported and locally produced cereals in the Namibian food security situat ion at nat ional and household levels, as stressed in a recent Discussion Paper (Fowler, 1997): « Among low-income households, food consumption (both cash expenditure and consumption in kind), especially of staple grains, accounts for over 50% of total consumption: in the northern regions, the purchase of mahangu, maize and other cereals by ‘low-income’ households accounted for 41% of total food expenditures, while among the ‘very low-income’ households the figure was 46%, compared to 24% amongst the ‘bet ter-off’ households. T hus changes in consumer prices of these grains have an immediate impact on the living standards, food security and nutrit ional status of the poorest sect ions of the population. »

CHAPTER 3 ON-FARM MAHANGU POST-HARVE ST SYST EMS This chapter examines the mahangu post-harvest systems in Northern Namibia at the level of rural households. Post-harvest operat ions include the following steps: harvest ing and drying, select ing next season’s plant ing seeds, threshing and winnowing, t ransport from the field and storage (intra- and inter-seasonal), home-processing and food ut ilisat ion, and market ing (including cash sale and bartering) of part of the household stock. It must be said that household post-harvest systems in Namibia have not been thoroughly researched and there is only fragmented literature available on the topic. The very likely intra- and inter-regional differences in the northern communal areas are also not well documented. Further research and development is required as a matter of priority for a bet ter understanding of post-harvest constraints faced by communal farmers, so that targeted solut ions can be promoted. Harvesting and Drying Farmers usually wait for mahangu heads to dry part ially in the field before harvest ing, unless the crop lodges seriously. Mahangu harvest ing generally takes place around May/June, dependin g on the rain pattern of the year and of the region, the earliness of the variety cult ivated, etc. The stem is cut with a sharp knife just beneath the head. T he heads are placed in large harvest ing baskets and taken away for further drying before threshing. It has also been reported that in Uukwaludhi (Om usati region) farmers generally cut the early crops at the base and stook the plants in the field before it is completely dry; the harvest ing of grain heads is done at a later stage (Matanyaire, 1998). Harvesting generally takes place at physiological maturity of the crop, which may not be reached uniformly amongst the plants in the field (and may require harvest ing in stages). At physiological maturity, the crop st ill contains a high percentage of moisture and is therefore vulnerable to damage during handling an d subsequent pest attack. This is why the crop is often left to dry in the field beyond physiological maturity. However, leaving the crop in the field too long may have negative consequences: higher losses to birds an d other field pests, increased insect infestat ion, etc. Harvest ing is therefore done as soon as the crop has lost enough moisture to enable safe and easy handling and before excessive field losses have taken place. The t iming of harvest is a compromise made by farmers balancing crop moisture and field loss, as well a s availabi lity of labour, risk of roaming grazing animals, etc. Straight after harvest ing mahangu heads are sun-dried, most often on a raised wooden platform near the fields and with a 24-hour watch against birds and other grain-eat ing animals. Drying can take a few weeks (to get the grain moisture content down to 10% or so). Drying on the threshing floor has also been reported, but the use of an elevated platform has the following advantages: contamination is minimised, with proper construct ion they can be made rodent-proof, and in the event of an end of season rain shower crop damage is minimised by quick re-drying due to the improved aerat ion (Unicef/MA WRD, 1999). To our knowledge, there is no report on research done in Namibia on grain drying prior

to threshing and no clear account of the drying parameters (durat ion, moisture content curve, etc.). T his had led CRIAA SA-DC consultants to recommend that drying of grain, and mahangu in part icular, prior to threshing be specifically researched (CRIAA SA-DC, 1999a & 1999c). Next Season’s Planting Seed Keeping seeds from selected mahangu heads for plant ing the next season is a small but crucial part of post-harvest systems. Farmers seem to choose the large heads from the best-looking plants in their fields. These mahangu heads are thereafter carefully stored apart from the rest of the harvest (most often in an unthreshed form) in the homestead. There is limited literature in Namibia on what criteria farmers use in their fields to select the mahangu heads they will keep as seed for the next season. Northern communal farmers manage a mixed ‘portfolio’ of short and long-cycle mahangu varieties, providing them with a desirable combination for spreading risks against unpredictable weather patterns and avoiding total crop failure (Matanyaire, 1998). Long-cycle variet ies (average of 120 days) are the t radit ional landrace cult ivars, selected by farmers over generat ions to preserve desired plant types and maintain genet ic diversity (or heterogeneity). The early maturing Okashana-1 (85 days), a cert ified seed variety with higher yield potent ial, has been rapidly adopted since its release in 1989. By the 1996/97 cropping season, this variety was so wn on an est imated 49% of the country’s mahangu area (Rohrbach & al., 1999). « Some farmers sow Okashana-1 at the beginning of the rainy season in order to obtain an early harvest, at the t ime when household grain stocks are low or exhausted altogether. (...) However, most farmers seem to be pursuin g a pract ice of sowing their long-durat ion tradit ional variet ies first (between October and December), with the early-season rains, and sowing Okashana-1 later in January-February. T his allows them to benefit from the yield advantages of their t raditional variet ies if rainfall is consistent through the cropping season. If the rains are late, a larger proport ion of land would be so wn to Okashana-1; if the rains are early, more land may be sown to tradit ional variet ies. » (Rohrbach & al., 1999) However, as Matanyaire (1998) pointed out: « (...) farmers had no intent ion of replacing all the landraces with Okashana-1. T hese findings provide clear signals farmers are more concerned about stable system productivity as opposed to yield maximisat ion, and that they recognise genet ic diversity as a viable option. » A study of Namibia’s present and future requirements for seeds (Balogun & Tripp, 1997) mentioned that because it is difficult for farmers to maintain the genet ic purity of short-durat ion mahangu variet ies, the annual commercial demand for certified seeds (such as Okashana-1) can be expected to be maintained over the years. Farmers generally cult ivate plots with a mixture of several variet ies of mahangu (an open-pollinated crop), which may all flower at the same t ime: « Such a crop management is not conducive to maintaining varietal purity of short durat ion variet ies. The problem of maintaining short-durat ion varieties is further

exacerbated by the pract ice of select ing large heads for seeds. Large heads are associated with long season length so farmers actually select against the main characterist ic they value in Okashana-1. » Two addit ional early maturing mahangu variet ies, Kangara and Okashana-2, were released in April 1998 (Rohrbach & al., 1999). T ogether with Okashana-1 (and a new white sorghum variety, Macia), these variet ies are mult iplied and sold a s cert ified seeds by the Northern Namibia Farmers Seed Grower s Co-operat ive (NNFSGC), based in Mahenene. T his means that northern communal farmers have access to a range of short-season variet ies on the market, although Okashana-2 and Kangara germplasms are close to Okashana-1 (Ipinge, 2000) but with some different traits (see below), and that farmers wil l continue their own seed select ion in the field, based on phenotypic characteristics. Farmers’ perceptions of the qualit ies and weaknesses of the improved mahangu variet ies were assessed by agricultural researchers (KFSRE, 1997; Rohrbach & al., 1999; Mukundu in MAWRD, 1999c). When farmers were asked about the main reasons for their interest in Okashana-1, the most common response was its early maturity, followed by grain yield and grain size. « Farmers perceive, at least , that this variety [Okashana-1] is more resistant to the range of different types of drought they commonly face - short seasons, limited rainfall, and mid-season dry spells » (Rohrbach & al., 1999) Farmers have cited three main weaknesses of Okashana-1 compared to local variet ies. The greatest problem is the weak stem, which causes plants to lodge at the end of the season (and reduces its value as buildin g and fencing material). T he second problem is the low stover yield (as livestock feed during the dry season). The third problem is the softness of the grain (which reduces its storability). Addit ional complaints are cited: the grey flour, which turns darker when cooked in porridge, and the tendency of the grain to shatter (the pericarp breaks into small pieces that are difficult to winnow) during dehull ing (Rohrbach & al., 1999). Farmers have, however, commented that Okashana-1 and -2 are easier to thresh (KFSRE, 1997; Mukundu in MAWRD, 1999c). Farmers’ ranking of grain t raits compared to the local variety was summarised in Rohrbach & al. (1999) and is reproduced in Table 4 belo w. Table 4. Farmers’ ranking of grain traits compared to the local variety (1993) Trait Okashana-1 Kangara Grain size Better: 90%; sam e: 10%; wors e: 0% Better: 70%; sam e: 30%; wors e: 0% Grain colour Better: 20%; sam e: 30%; wors e: 50% Better: 80%; sam e: 20%; wors e: 0% Dehulling quality Better: 100%; same: 0%; wors e: 0% Better: 100%; sam e: 0%; wors e: 0% Grinding quality Better: 0%; sam e: 0%; worse: 100% Better: 100%; sam e: 0%; wors e: 0% Food preparation Better: 100%; same: 0%; wors e: 0% Better: 100%; sam e: 0%; wors e: 0% Taste Better: 20%; sam e: 40%; wors e: 40% Better: 0%; same: 100%; wors e: 0% Source: Ipinge & al. (in ICRISAT, 1996). Threshing and Winnowing Threshing is the detachment of kernels from the rest of the head (or panicle). For

mahangu it also involves removing the chaff, which is carried out by winnowin g, to obtain a clean grain. Threshing is very important in the small-holder post-harvest operat ions for on-farm grain storage and market ing. Properly threshed and cleaned grain is also crit ical to the t rading and processing sector. Threshing is very often a work bott leneck in the small-holder farming calendar. Mechanisation of threshing is considered important not only for alleviating these t ime and work constraints, but also to reduce post-harvest grain infestat ion by pests, as well as grain contamination by sand, dust and micro-organisms. To our knowledge, there has been no comprehensive study of mahangu (and other cereals) threshing in northern Namibia. The following paragraphs are an attempt to summarise what is known about it and what has been done to promote improved methods. Pre-storage threshing In northern Namibia, cereals and mahangu in part icular are stored in grain form at the homestead. T hreshing is therefore done before storage of the crop. Threshing is most often carried out in or near the crop fields at the drying place, which is often enclosed with wooden poles and/or thorny bush branches. Straight after threshing, the grain is brought to the homestead for storing. Depending on when the rain stops and how long harvest ing-drying requires, threshing generally starts around June. It can last unt il July/August in a bad harvest year, but it can continue into September/October in an especially good year when mechanised threshing is not available and labour is scarce. Manua l threshing and winnowing are typically the work of household women, but men and children often join the family working team. It is also not uncommon that labour is hired to speed up the threshing operat ion. There is limited data accounting for the t ime required by manual threshing (and other post-harvest operat ions) in the overall mahangu crop production of northern communal farmers. Studies account for labour t ime required per hectare for ploughing, plant ing, weeding an d harvest ing/threshing, but it is never clear what exact operat ions are included in the latter. Although labour requirements are unclear in non-mechanised mahangu crop production systems, post-harvest operat ions and threshing in part icular are considered work bott lenecks in a year of above-average harvest . T his is so because threshing is labo ur intensive and competes with other on-farm and off-farm act ivit ies, and above all because it is desirable to bring back the threshed grain to the homestead as soon as possible to minimise the risk of loss and relieve scarce labour from guarding the unthreshed crop. The different methods of threshing used in the Northern Communal Areas The ‘tradit ional’ way to thresh is by usin g wooden sticks to beat the mahangu heads on the ground. In the NCRs, the heavy pounding pole (pest le), fitted with a cylindrical wooden piece at the end, is inverted and used for threshing. Otherwise, a straight wooden st ick or pole is used. T here are no accounts of flails or improved threshing frames being used. Mahangu heads are beaten direct ly on hardened and cleaned groun d, sometimes on a compacted soil platform, or on a tarpaulin, or in a bag. Winnowing is carried out thereafter with special flat baskets, but also with other types of containers such as plast ic dishes and galvanised basins, assisted by the breeze.

Dendy (1993) reported that the productivity of manual mahangu threshing (including winno wing) measured in Kavango (usin g light poles) was 5 to 10 kg of threshed grain per women-hour (5 to 10 women-hours per 50 kg bag). On three runs, the average ‘out-turn’ (weight of grain/weight of mahangu heads) was 77% (ranging from 70% to 83%). Winnowing represented between 34% and 40% of the total woman-time of threshing. Similar rates, ranging between 4.5 kg and 11 kg/hour, were observed for mahangu in Caprivi (CRIAA, 1996). ‘Tractor t reading’ is another method used. Panicles are gathered on the ground in a ring or a heap over which a t ractor runs. T he rolling tractor tyres execute the threshing operat ion (to avoid breaking the grain, tyres are deflated or lighter t ractors are used). People stand around the heap and push back the mahangu heads drawn aside. Winnowing must be carried out thereafter (CRIAA SA-DC, 1999a). This method seems to be used much more, especially in the NCRs, than one would have thought . A tractor service consultancy (NRC, 1998) mentioned that 82% of the t ractor entrepreneurs interviewed in the NCRs (97 in total) provided this form of threshing service. Farmers were charged per heap, but unfortunately no est imate of the weight of a heap was provided. The performance of threshing by tractor t rampling is not known and should be studied. Manual or t ractor threshing of mahangu heads direct ly on the ground results in sand (and dust) contamination of the grain. It is not easy to subsequently separate the sand from the grain and this has serious implicat ions for commercial processing requirements. T he advantage of mechanical threshers, beyond the increased productivity of operat ion, is that they have an in-built winnowing system, and discharge the clean grain direct ly to bags or baskets. Mechanical threshing appears to be mainly performed on a service basis by ur ban and rural entrepreneurs, who are farmers themselves. T here are two models of mahangu thresher most commonly used (CRIAA SA-DC, 1999a): The ROVI C mahangu thresher is so far the only locally manufactured machine. It is a relat ively light thresher run by a 3 hp petrol engine, which can be t ransported on a pick-up vehicle, or on a cart or a sledge. T he manufacturer indicates a capacity of 400 kg per hour (confirmed by Dendy, 1993, on two runs of 10 minutes) but field data on longer periods of use suggest a lower throughput around 100 kg to 200 kg of threshed grain per hour. Some other technical performance and limitat ions have been reported, but because it is the only portable thresher manufactured and available in Namibia, its use has been spreading in the NCAs. The much larger t ractor-PT O driven thresher mounted on four wheels, with a capacity of around 1 tonne per hour, is better suited to farms with large mahangu fields and lar ge harvests. These machines were originally designed for maize shelling and are converted to mahangu threshing with some adaptat ions, part icularly by subst itut ing a drum with smaller perforations. All these threshers are second-hand machines, imported from South Africa by local farm equipment dealers (who recondit ion and adapt them). A technical field assessment of these two threshers has been recommended (CRIAA SA-DC, 1999a & 1999c). Economic overview of threshing

It is not known what proport ion of the annual mahangu harvest is threshed according to which method and how many farmers use which threshing technology. It is also not known how many mechanical threshers (and what models) are in operat ion in the northern communal areas. It is therefore impossible to accurately evaluate the present economic importance of mechanical threshing versus manual and tractor t reading threshing. A hypothet ical economic calculat ion was, however, at tempted and it is presented in Box 1 below (CRIAA SA-DC, 1999a). Box 1 Economic Overview of Mahangu Threshing in NCAs We consider that roughly 25 % of the mahangu harvest is mechanically threshed and we arbitrarily set a future target for mechanical threshing of 75 % of the nat ional mahangu harvest . With this opt imist ic target , we assume the following: • T he annual total mahangu harvest is between 25 000 t and 100 000 t depending on the

year, and the corresponding mechanical threshing targets are 18 750 t and 75 000 t ; • In bad years, threshing lasts 1 month (25 working days) and in good years, 4 months

(100 working days); • T he Rovic machine threshes an average of 1 t per day (25 t per month, 100 t per 4

months); • T he PT O-driven machine threshes an average of 8 t per day (200 t /month, 800 t /4

months); • T here will be one PT O thresher for eight Rovic machines. With this target, the future need of threshing equipment in working order will be 374 Rovic threshers and 47 PT O-driven machines, representing an investment cost of some N$ 2 million. At an average threshing charge of N$ 100 per tonne (a low N$ 5 per 50 kg bag), the annual turnover of the threshing industry (covering 75% of the needs) could reach between N$ 1.87 million and N$ 7.5 million, depending on the year. With this mechanical threshing target of 75 % of the mahangu harvest as compared to 25 % est imated for the present , manual threshing t ime savings (at an average rate of 100 kg per work-day) wo uld represent between 125 000 and 500 000 work-days per annum, which is considerable. In monetary terms, with a work-day valued at N$ 15, the t ime saved by the further mechanisat ion of threshing would represent a value of between N$ 1.9 million and N$ 7.5 million per year. Legit imate concerns have been raised abo ut the socio-economic impact of mahangu threshing mechanisat ion on manual work opportunit ies for low income people and especially women in a rural economy with limited job opportunit ies. Experience elsewhere in Africa shows that, on the contrary, the time freed by mechanisat ion is used for other cash generat ing act ivit ies. However, in the NCRs rural economy where the opportunity cost of labour is relat ively low, manual threshing and mechanised threshing have a good chance to co-exist. • At an opportunity cost of N$ 15 per day, manual threshing costs N$ 150 per tonne

representing 15 % of the price of mahangu grain (for a market value of N$ 1 000 /t ), which is very high;

• At N$ 10 per day, manual threshing costs N$ 100 per tonne, which could be at par with mechanised threshing service charges;

• At N$ 5 per day (N$ 50 /t ), manual threshing would out-compete mechanical threshing on price;

It has been generally considered that mechanised threshing costs should be kept below 10 % of the market price of grain. It is concluded here that mechanised threshing would expand if its cost remains around or below 10% of the market price of grain, which wo uld require efficient machinery and good management of threshing services. In view of the limited range of threshing models available, the operat ional constraints experienced by farmers, threshing service entrepreneurs, traders and processors, and a lack of precise data on most technical and economic aspects of threshing, specific research in the form of a baseline survey was recommended to take place as soon as possible durin g the forthcoming threshing season. Terms of reference were drafted (CRIAA SA-DC, 1999a) and comprised the following elements: Technical field assessment of the different methods of drying and threshing in the NCAs; Technological search for appropriate threshing (winnowing and pre-cleaning) technologies from the SADC region and elsewhere, to be imported or possibly locally manufactured (locally adapted or at least locally assembled); Financial and operat ional assessment of service threshing in NCAs; Defining a promotion strategy for improved mechanical threshing in NCAs. On-Farm Storage In a rain-fed cropping system based on one annual production cycle over a short rainy season, the harvest obviously needs to be stored for household consumption through the year. When rain pat terns are unpredictable and ‘droughts’ can occur on a recurrent basis, as in northern Namibia, inter-annual storage of the staple cereal is a requirement for household food security. From his field survey conducted in 1992/93 in the North Central and Kavango regions, Keyler (1996) reported on the effect that fear of drought had on household mahangu reserve levels (and decisions to market surplus). On average, farmers’ fear in the NCRs translated into ‘almost unrealist ic’ mahangu reserve targets of between 4 and 6½ years, meaning that household reserves had priority over commercial sales. Kavango households that never or only rarely sold millet preferred to have, on average, mahangu reserves that last 2 years. Kavango households that sold mahangu every year perceived a need to set their reserves at , on average, 3½ years. Matanyaire (undated) confirmed the regional differences between the NCRs and Kavango in the way farmers managed their inter-annual mahangu reserves. The proport ion of households opting for storing their mahangu in excess of annual food requirements was higher in the NCRs (81%) than in Kavango (54%). T he majority of households (65%) also indicated that mahangu could be stored without losing ‘food qua lity’ for three or more years. Northern communal farmers store threshed grain either in granaries in the homestead, or inside the home in different types of containers (bags, baskets, drums ...). Compared to

the North Central Regions, t radit ional mahangu storage is much less documented in Caprivi and Kavango, where a variety of storage containers and structures are used, including granaries made of earth blocks or poles and mud, raised on a low platform and roofed with thatch. « In a field study, Keyler (1995) revealed that in the NCRs 90% of the households stored their pearl millet grain in t raditional storage baskets, while in Kavango the largest group (29%) kept it in jute or polypropylene bags, 21% used sealed dr ums and another 19% kept the bulk grain in storage huts. » (Matanyaire, 1998) In the North Central Regions, the t radit ional granary consists of a large spherical woven ‘basket’ (or ‘bin’) made of Mopane (Colophospermun mopane) branches and interlaced with strips of Mopane bark. The interior of the bin is sealed with a light coat of ‘mud’, which is often made of termite hill soil mixed with water. T he bin has a circular opening on top, through which the grain is loaded and removed, and is closed with a lid (which can be sealed with mud). T he bin is raised on a wooden-legged circular cradle. A thatched roof, under which the bin is placed, is the last part of the granary. Bins vary in size from 0.3 m³ (holding approximately 250 kg of grain) to over 2 m diameter structures (NRI, 1997), that is around 4 m³, holding over 3 tonnes of grain. There is anecdotal evidence in the North Central regions of innovative grain storage containers being made by farmers: small concrete ‘silos’ (similar to a water reservoir) and thick baskets with lids made of Makalani palm leaves (Hyphaene petersiana). In the Shambyu area of Kavango (Mupapama river terrace community), two types of granary have been reported and photographed (Eirola & Bradley, 1990). T he first type is made of poles and plastered with clay, raised on a wooden platform (some 60 cm above the ground) and thatched like a t radit ional hut. A small entrance leads to the inside. T hese granaries are located outside the homestead enclosure. T he second type consists of large grain ‘baskets’ made of split river reeds (as sleeping mats are made), raised on a wooden floor and covered by a thatch roof. T hese storage baskets vary in size with an average capacity of about 400 kg of grain. T hey are located inside the homestead. In the same area, a variant of the first type was captured on a picture (Yaron & al, 1992). The inside of the granary has two compartments, which are accessed from two small openings in the wall. T here is no clear account on the ‘ownership’ and management of household grain reserves, apart of the following report from Kavango: « Just as the husband and wife or wives have their own fields, they also have their own mahangu storage places. T he wife’s or wives’ mahangu supplies are used first and only then the husband’s. » (Eirola & Bradley, 1990) Households of the NCRs o wn one or more granaries according to the size of their fields and quantity of mahangu produced. When a family has only one granary, it is filled with the last harvest . If grain from the preceding year remains, it is placed on top of the new grain or placed in another container. But if a family has several granaries, the different harvests are kept separate. If stored mahangu had not been completely consumed by the time of the new harvest, families most often prefer to finish the old harvest stock, because new grain is m uch harder to pound (Labetoulle, 2000). Farmers in both North Central and Kavango regions reported varying degrees of storage

losses (Key ler, 1996; Matanyaire, 1998). As Hindmarsh (2000) noted: « On-farm storage losses for mahangu appear to be unusually high due to pract ise of storing the grain threshed and retaining surpluses on farm to provide about two years’ supplies. Since the oldest crop is eaten first , insect infestat ions have two years to develop in storage. » Nevertheless, a third of Kavango farmers and 82% of NCRs farmers were taking measures against insect infestat ions in their grain, roughly 12% of farmers in both zones were using chemical insect icides, while the rest of the farmers were usin g tradit ional methods like ash or leaves (Keyler, 1996). T hese t radit ional methods are not well documented. Mopane wood ash seems to be the most commonly used. T he ash is put in layers in the grain when the granary or the container is loaded, or the ash is spread at the top on the grain when flying insects appear. The most significant mahangu storage pest causing losses is reported to be Corcyra cephalonica. « T hese moth infestat ions result in masses of grain held together by webbing (silk) produced by the larvae as they move through the grain seeking a pupation site. Many individual grains have their embryos removed by the feeding larvae. In order to use the grain, they have to be rubbed an d sieved to remove the webbing, or alternatively the masses of clumped grain are fed to chickens. » (NRI, 1997) On-farm physical losses in grain we ight have not been assessed, but were crudely est imated to range from 10% after one storage year to more than 30% over longer storage periods (NRI, 1997). Farmers, especially in the NCRs, have complained about the poor storability of the improved variet ies, such as Okashana-1 (Matanyaire, 1998). Grains with large, soft endosperms and thinner pericarps - characterist ics of the improved variet ies released in Namibia - are more susceptible to insect damage (especially by feeding moth caterpillars) than the small, hard landrace types. On-stat ion trials conducted during the 1999-2000 storage period by the FSRE Unit in collaborat ion with NRI proved that infestat ions by the moth Corcyra cephalonica actually started in the field before the grain was bro ught to storage. It was concluded (Hindmarsh, 2000): « Physical barriers surroun ding the content of the storage container are not effect ive, since infestat ions are act ive on the grain before storage. We can assume that the infestat ion would not have been localised at the surface when the bins were filled (because of thorough mixing of samples). Therefore, it is safe to conclude that C. cephalonica larvae migrate vert ically to the surface of the grain bulk to pupate. Discussions between researchers and farmers invited to the init ial opening of the containers led to the proposal that a layer of wood ash placed near the top of the grain bulk would intercept migrat ing larvae and could prevent pupation and emergence as sexually act ive adults. (...) Such a t reatment [wood ash] could be cheap and very effect ive, and it is likely to be acceptable to farmers. Sieving and washing to remove ash residues are normal pract ice for households before the grain is pounded into a meal. » The price of Mopane storage baskets has become relat ively high and a shortage of local materials to build them has been reported. NRI (1997) recommended developing, test ing

and promoting cheaper and improved on-farm storage structures, galvanised iron bins, fibreglass and plast ic containers, which had proved viable in other countries. This was also recommended by Hindmarsh (2000). The Rural Development Centre (RDC) in Ongwediva has embarked on manufacturing and selling storage containers made of corrugated metal sheets. The RDC recently reported having sold 68 units over the past three years or so (F. Msati, pers. comm.), but in the absence of field monitoring it is not possible to assess the impact of this technology on household grain storage. Household Food Processing The presentat ion below dist inguishes between primary processing - processing of the whole grain into a consumable product to be cooked - and secondary processing, which refers to further processing and/or food preparat ion. It is by no means a comprehensive presentat ion. It ignores inter- and intra-regional differences and the diversity of individual recipes at households level. It can be taken, however, as a summarised account of the ‘t radit ional way’ in food processing, with a special focus on the North Central Regions, which has been documented more extensively (especially in Labetoulle, 2000). Mahangu primary processing Mahangu grain is processed into flour in two stages: first dehulling, then pounding into flour (pulverising or size reduction). In northern Namibia, mahangu processing is mainly done manually, using wooden mortar and pest le. Dehulling separates the bran, consist ing of the outside envelopes (pericarp, testa ...) and a port ion of the germ, from the endosperm (mainly consist ing of starch). T he average composit ion of pearl millet grain is shown in Table 5 below. Table 5. Average composition of pearl millet grains (percentage o f total weight) Large grain Mediu m grain Small grain Whole grain 100 100 100 Endosperm 76,0 75,0 74,0 Germ 17,0 17,5 15,5 Pericarp 7,0 7,5 10,5

Adapted f rom Abrasive-Disk De hullers in Africa (1989) Millet and sorghum contain nutrit ional inhibitors, the level of which has to be reduced before consumption. The presence of polyphenols (or tannins) gives the grain its bitter taste and reduces its digest ibility. In the field, the bitterness makes the grain less att ract ive to birds and reduces pre-harvest losses. Phytic acid is also found in millet and sorghum grain, mainly in the aleurone layer (a thin layer rich in protein and fat found between the endosperm and the pericarp). Phytic acid has ant i-nutritional effects on minerals and affects the digestibility of proteins and starch. It can be reduced significantly by dehullin g. Tannin and phytic acid bind minerals and proteins, preventing absorpt ion into the body: « T o reduce the undesirable tastes due to the presence of polyphenols in sorghum and millet , the rural housewife dehulls the grains and often allows the processed product to ferment. T hese processing techniques - dehulling and fermentat ion - lower the

polyphenol levels in the grains, In other cases, however, where grains with high polyphenol levels are unacceptable for food preparation, they are fermented into beer. (...) Phytic acid forms insoluble compounds (phytates) with such mineral elements as calcium, iron, magnesium, sodium, and zinc, making them unavailable for use by humans. Phytic acid can also combine strongly with proteins at certain pH levels. » (Bassey & Schmidt, 1989) Grain deh ullin g, by separat ing the endosperm from the outer envelopes: Reduces the fibre content and improves the taste, texture and nutrit ional value of the finished product; Removes the bitterness (tannins) that is found mainly in the pericarp and the testa, the layer situated immediately below the pericarp. Both dehulling and poundin g into flour can be mechanised and different models of machines exist . However, their use by rural families is not widespread in northern Namibia. T o our knowledge there is no hand-operated grain grinding machine being used at household level in the NCAs. Rural families use mechanised service mills when they have cash available and especially when there is too much work in the fields, and if the mills are not located too far away. When women bring their mahangu grain to a service mill, it is in the form of home-dehulled grain, which has very often been soaked and fermented, and is st ill wet . Milling of wet materials has technical and economic implications for mills with equipment designed to process dry cereal (see next Chapter). Dendy (1993) reported that Okashana-1 as compared to local mahangu variet ies has the following processing features: It is easier to dehull (due to slight ly looser hull); It is harder to pound; It takes longer to soak (to soften the grain before pounding); Overall, it takes longer to process into meal and the out-turn is lower. The comparat ive physical characteristics of Okashana-1 compared with different tradit ional variet ies are provided in Table 6 below. Table 6. Physical characteristics of Okashana-1 and local mahangu varieties (Kavango) Okashana 1 Large grain (local) Small grain (local) Bulk density 1 250 ml /kg 1 225 ml / kg 1 333 ml / kg Thousand grain weight 11.2 ± 0.5 g 9.4 ± 1.8 g 4.5 ± 0.2 g

Source: De nd y, 1993 Okashana-1 and the recently released mahangu varieties had the following characterist ics (MA WRD, 1998; Rohrbach, 2000): Okashana-1: large grain, grey to grey-yellow in colour, milling yield of 81.4% - 86.2%; Okashana-2 (Kakunya): medium panicle length (18-25 cm), large grain (13.3 g per 1000 grain weight), grey to dark grey in colour, milling yield of 82.3% - 93.4%;

Kangara: medium panicle length (16-22 cm), large grain (over 12 g per 1000 grain weight), cream yellow in colour, milling yield of 84.5% - 92.0%. Measurements of hand-pounding rates in Kavango (Dendy, 1993) showed that in household preparat ion of meals, production is only 700 g/women-hour for local mahangu variet ies and 600 g for Okashana (see Table 7). T he household out-turn of meal is about 50%, much lower than for machine milling: « A surprising conclusion from these results is the measurement of the loss as bran and ‘middlings’. T he bran [envelops and part of the germ] makes up approximately 15% by weight of the grain. By hand-pounding, some 30% of the endosperm (that part which is desired for food and is around 80%-90% of a grain) is being “ lost” in the bran. T his is used for brewing and, eventually, as feed. However, in terms of food value, this is a very high loss compared to the optimum machine out-turns of over 80% - i.e. 20% bran. » Table 7. Comparison of Okashana-1 with local variety by hand-pounding (on 6 runs) Okashana 1 Local vari ety Stage 1: dehulling (woman mn / kg grain) 19.4 20.6 Stage 2: pounding to meal (woman mn / kg grain) 29.1 21.7 % dehull ed 72.0 72.8 % meal obtained 49.1 54.6 Rate o f p roduction: kg meal / woman-hour 0.60 0.70

Source: De nd y, 1993 The above dehulling productivity of around 3 kg per hour seems low compared to the operat ions seen at Kamalanga’s Mill in Kavango, where women were manually dehull ing Okashana-1 grain (piece-work for cash) at a rate of 6 kg to 9.5 kg per hour, over a 9 hour work-day (CRIAA SA-DC, 1999a). The overall hand-pounding productivity of around 0.6 - 0.7 kg of meal per hour seems also low compared to the figures provided belo w, but could be explained by the high level of loss in the bran reported by Dendy. Field survey reports suggest that families do not prepare mahangu flour every day, but rather once, twice, or thrice a week (or sometimes more). T he frequency depends on the season (agricultural calen dar), on the availability of home reserves of mahangu, and on the number of persons at home (Kroll, 1991; Pitois & Helmstetter, 1996; Labetoulle, 2000). In Uukwaluudhi (Omusati region), Kroll (1991) reported a processing rate of 8 kg of mahangu grain per hour involving 3 women (that is around 2.7 kg per woman-hour). For an average family size of 9 (4 adults and 5 children), 3 women were each working close to 13 hours per week on mahangu processing (a total of 39 women-hours per week). T he average mahangu grain consumption was 37 kg per week, equivalent to 4 kg per person per week, and to 570 g per person per day. Pitois & Helmstetter (1996) reported very similar processing and consumption figures in Onamutanda (Ohangwena region). These data would translate into an annual consumption of over 200 kg of mahangu grain per capita, which is much above the figures given by the nat ional consumption stat ist ics (110 kg - 135 kg of cereals). It is plausible that the level of loss in manual processing, as

reported by Dendy, would explain the difference, i.e. an annual consumption of some 200 kg of grain represents an actual consumption of about half in flour form. But the nat ional stat ist ics on mahangu and other grain ut ilisat ion might be underest imated, which would concur with the opinion that nat ional production stat ist ics are also underest imated, as presented in the previous chapter. It is nevertheless clear that hand-pounding mahangu is one of the major burdens of women in rural areas. In Onamutanda, it was reported was 44% of surveyed households were hiring people to pound their mahangu and the average payment was around N$ 3.00 per lata (or N$ 0.20 per kg) in 1995 (Pitois & Helmstetter, 1996). The tradit ional preparat ion of mahangu flour follows a num ber of steps, which lengthens the process and would make it difficult to mechanise ent irely. The tradit ional slightly fermented mahangu flour is, however, the product that receives overwhelming preference from mahangu consumers. T he main steps, from the whole grain to the final flour (or meal) ready to be cooked, are presented in Table 8 below with their respective by-products. Table 8. Traditional Preparation of Mahangu Flour Processing step Method/tools Produc t B y-product Use of by-product 1. Winnowing and screening grain

Flat winnowing basket and by hand

Clean whole grain Foreign matter + badly damaged grain, insects and webbing

Waste + poultry fe ed

2. Dehulling Wetted grain pounded in mortar, then dried and winnowed

Dehulled grain Yellowish bran + pounding loss and winnowing dust

Traditional drinks + poultry feed

3. Fermenting Soaking in water (few hours to 1 day) + already fermented pounded grain added as starter (from 5.), then trickled and sun-dried

Fermented dehulled grain (whiter and tastier)

Waste liquid Thrown away or given to pigs (other use? )

4. First pounding into flour + sieving

Mortar and pestle + hand sieve

First white flour + grits to be pounded again

P ounding loss P oultry feed

5. Second pounding + sieving

Mortar and pestle + hand sieve

Second white flour and some partly pounded grain kept for ne xt day(s) fermenting (to 3.)

P ounding loss + re maining bran and germs

P oultry feed

6. Drying of flour

In flat basket or on cloth in the sun

Dried fermented flour to be cooked or stored (few days)

(Loss by wind)

Source: Mallet M. & Amutse F. (CRIAA SA-DC, unpublished) Soaking and fermentation of grain before final grinding to flour is a t radit ional pract ice to improve digest ibility and nutrit ional value of grain products, improve taste and texture and to extend the shelf life of flour: « Soaking is the init ial step of malt ing and fermentat ion. The common t ime of soaking grain is 24-48 hours in lukewarm water in tradit ional food processing. Soaking of grain in

water for 24 hours before milling is enough to increase the amount of soluble iron up to 10 t imes. During the t ime of soaking, the lactic acid fermentat ion is developing and the pH drops. (...) The changes occurring during the fermentat ion process are mainly due to enzymatic act ivity brought about by the micro-organisms (yeast and bacteria) and/or the indigenous enzymes in the grain. T he two major types of bacteria in cereal fermentat ion are lact ic acid- and acetic acid-producing bacteria. T he micro-organisms involved in natural fermentat ion of cereals are essentially the surface flora of the seeds. During fermentat ion the pH drops from about 6.5 to 3.6. This effect ively inhibits the growth of other bacteria that cause decomposit ion and food spoilage. They will also have a strong inhibit ional effect on diarrhoea-causing pathogens (...). Lact ic bacteria are reported to contain proteolyt ic act ivity capable of degrading more complex proteins into simple proteins, peptides and amino acids. T his might explain the favourable effect on protein digest ibility found after lact ic fermentat ion of cereal grains. Fermentat ion of cereals inhibits the negat ive effects of phytic acids and tannin. It is evident that fermentat ion is an effect ive and low-cost method of preventing/reducin g diseases caused by deficiency of important nutrients in regions where cereals form a major part of the diet . » (Jonsson & al., 1994) Mahangu secondary processing and food preparation Mahangu is mainly consumed in st iff porridge and fermented drinks, but there are other food preparat ions, which have been documented for the North Central Regions (Labetoulle, 2000). In summary these basic food preparat ions are as follows (with their Oshindonga names in italic): St iff porridge (Oshithima or Oshifima in Oshikwanyama): mahangu flour cooked in water. Pancakes or unleavened bread (‘Owambo bread’, Oshikwiila): mahangu flour mixed with water and seasoned with salt and/or sugar, moulded in lar ge flat tened balls, are either pan-fried, roasted on coals or boiled in water (the balls being put in plast ic bags). Uncooked flour mixed with sour milk (Olumbololo). Boiled grain, most often dehulled (Onona). Thin porridge or gruel (Etete): prepared the same way as the st iff porridge with less flour. It must be noted that rolled products, such as “couscous”, are not tradit ional in Namibia. Mahangu (and sorgh um) form part of the composit ion of various t radit ional fermented beverages. Sorgh um is used in malted flour form, while mahangu is very rarely malted in the NCRs. T hese home-prepared drinks are very popular in North Central Namibia and are consumed by a large part of the populat ion. These are: Ontaku (or Oshikundu in Oshikwanyama): a non-alcoholic fermented drink (lact ic fermentat ion) made from malted red sorghum flour, mahangu bran and mahangu flour. This drink maintains an important place is the diet of the NCRs populat ion, where it is consumed daily by adults and children alike. The opaque beer (Omalovu) made from malted red sorghum flour and mahangu flour. In addit ion, the t radit ional Tombo (made of malted red sorghum and brown sugar) and

Epwaka (made of mahangu bran and bro wn sugar) are much more fermented and alcoholic drinks, which are either prepared for special social occasions or marketed in ‘Cuca shops’. Malt ing is a process durin g which the whole grain is soaked and then germinated: « During this process a number of enzymes includin g phytases are act ivated provided the pH is kept within 4.5-5.5. At this pH level, ground malted cereals wil l have almost all phytic acid degraded. (...) Germination, the start of the growth of the seed to a plant, releases other enzymes from the germ in addit ion to phytase. The germinated grain is called malt and is rich in enzymes breaking down the starchy, white part of the grain to sugar. If yeast is present , either by addit ion or nat ive in the grain, then water is added to the ground malt and the sugar wil l break do wn to alcohol. T his is the principle of beer making (...). Ground malt which contains the enzyme amylase is also used in porridge to increase nutrit ional value of baby food by increasing the actual nutrit ion per spoonful and to make desirable and healthy drinks (...). » (Jonsson & al., 1994) In the NCRs, mahangu is generally used as the main ingredient in weaning foods, most often in a specially prepared Ontaku, or in a thin porridge (with or without sour milk and possibly sweetened with sugar) (Labetoulle, 2000). As most rural households in the NCAs are not self-sufficient in mahangu, maize meal is purchased and consumed in porridge either as a substitute for mahangu or mixed with it . Some families prefer to buy mahangu grain rather than maize meal, and some others buy maize meal even when their mahangu grain stock is sufficient , to save t ime and labour (Labetoulle, 2000). Rohrbach (ICRISAT, 1995) reported on a survey (from the Namibia Millet Subsector Project) on households grain buying in December 1992 (seven months after the 1992 harvest in a severe drought year), which gives indications of the importance of maize purchase. T he two tables are reproduced below. Table 9. Proportion of households buying grain, Dec. 1992 Regions Buying grain Buying

mahangu grain Buying mahangu meal

Buying mai ze grain

Buying mai ze meal

NCRs 73.0% 12.5% 6.0% 14.0% 56.5% Kavango 83.3% 39.2% 13.4% 24.2% 66.7% Source: ICRISAT, 1995 Table 10. Proportion of grain purchased, Dec. 1992 Regions Mahangu grain Mahangu meal Mai ze grain Mai ze meal Sorghum grain NCRs 12.8% 4.4% 16.8% 64.2% 1.7% Kavango 28.1% 5.0% 12.1% 52.6% 2.0% Source: ICRISAT, 1995 Because the price of maize meal generally compared well with the cost of mahangu grain plus millin g service, it was concluded that poorer households tended to rely on maize meal to resolve their production deficits in both good and poor years. Only wealthier

wage earners were willing to pay the premium price for mahangu (Rohrbach & al., 1999). Household Grain Marketing Since the study carried out by Stefan Keyler during the 1992/93 production year, there has been very limited field-work conducted on mahangu market ing at household level. Some significant results from Keyler’s work are recalled below and are assumed to st ill be relevant to the present situat ion. It is not clear what significant socio-economic changes may have happened at household level since then, and how commercial mahangu milling and market ing facilitat ion intervention may have impacted on mahangu grain supply from farmers in the NCAs. Keyler (in ICRISAT, 1995) estimated the volume of Namibian mahangu traded in 1993 at 7 200 tonnes (excluding 1 700 t imported from Angola), virtually all in unprocessed form. It was further noted that older grain surpluses were offered to the market only when the new harvest arrived and storage containers became scarce. The main reason for selling mahangu was given as the urgent need for money and goods. A large proport ion of households only sold mahangu after a good harvest . Most of it is sold within village areas. After neighbours, the next largest category of buyer was t ravelling traders. Only a small proport ion was t raded through markets. « T he most important finding about millet market ing on the household level is that about 70% of farmers in Ovambo (and 28% in Kavango) never sell millet , while the remaining sell millet only in small quantit ies every other year. Most of the farmers that market millet sell it informally to their neighbour. » (Keyler, 1996) Keyler further discussed mahangu selling l imitat ions (in ICRISAT, 1995): « T he majority of rural households indicated that their millet t rade is mainly limited by the amount of grain they produced and the fear of drought . However, market ing constraints were addit ionally mentioned by those households that sold millet every year or at least in some years. Larger port ions of millet selling households of both study zones [NCRs and Kavango] mentioned long trading distances, the lack of t ransport , and high cost of hired transport also as limit ing factors. In Kavango the lack of local markets was an important constraint among those households that sold millet every year. » Sullivan (in ICRISAT, 1995) summarised the situat ion as it was understood at the t ime. It is very likely that it st ill applies to the present situat ion. « T he small quantit ies coming on the local markets do not necessarily come from farmers producing surpluses, but rather from producers with an immediate requirement for cash for the purchase of consumer goods or school fees etc. The millet sold is subst ituted for, at a later stage mainly, with maize meal as the product most readily available. (...) A relat ively small number of farmers, perhaps no more than 3 to 4%, produce a surplus of millet for sale. Even in good growin g seasons such surpluses range up to 5 tonnes while a few farmers produce up to 25 tonnes or more. T hese farmers are at the forefront in demanding a formal market for their crop. » Disposing of mahangu grain in small quantity to traders, relat ives or neighbours - for cash, as gift or in barter - is st ill common in the NCAs. In a study of four villages in the

NCRs (Labetoulle, 2000), it was reported that bartering was widely used to procure foodstuffs. Mahangu grain was the principal means of exchange, followed by sorghum and other agricultural produce. T he most common barter t ransact ions were mahangu for meat or fish, and mahangu for animal fat or clothes. Concluding Remarks On-farm mahangu post-harvest pract ices are extremely important for household food security of the rural populat ion in northern Namibia. Mahangu post-harvest issues are also critical to the development of a mahangu commercial sector, as the mahangu surplus that is t raded mainly originates from rural households (although only a small proport ion of them market mahangu regularly). Improving the quality of grain stored on-farm is important for the t rading and processing industry, which has very often complained about the unsuitable quality of mahangu grain marketed by farmers. As we have seen in this chapter, a number of issues are not well documented and warrant further applied field research: Mahangu crop drying after harvest . Seed select ion and storage of mahangu for re-plant ing the next season. Threshing methods used by farmers. Storage and other post-harvest losses. Nutrit ional aspects of fermented mahangu (flour and beverages) processing. Diversity and richness of traditional food preparat ion recipes (especially in Kavango and Caprivi regions, which are not well documented), consumer preferences, etc. Farmers’ post-harvest constraints that need to be addressed by research and development (R&D), can be summarised as follows: Table 11. R&D needs fo r addressing fa rmers’ post-harvest constraints Farmers ’ Post-Harves t Cons traints Research and Development Needs Long and labour intensive post-harvest practices

Need fo r mechanising the operations

Poor thres hing Need fo r improved threshing m ethods, includi ng mechanised threshi ng service

High price o f traditional storage bins and shortage o f m aterials fo r thei r const ruction

Need fo r improved and cheaper storage containers

Poor quality o f grain aft er long storage Need fo r reducing storage and post-harvest loss es (quantity and quality)

Poor quality o f grain m arketed Need fo r pre-cleaning technology and price premium fo r quality grain

Not enough grain storage capacity for surplus producers

Need to develop improved bulk storage and/or rapid access to market

No easy access to market Need to develop decentralised trading network for qui ck cash intake

No ext ension servi ces on post -harvest issues

Need to develop extension mess ages from applied R&D

Mahangu p rocessing labour intensive Need to promot e home milling technologies, and develop decent ralis ed service milling (including dehulling)

Traditional wet processing o f mahangu not adapted to available mechanis ed milling

Need to adapt improved/m echanis ed processing appropri ate for the p reparation of fermented m ahangu flou r

CHAPTER 4 GRAIN T RADING AND PROCE SSING SE CT OR This chapter looks at the procurement and ut ilisat ion of mahangu from the perspective of the t rading and processing sector. It describes the state of development of the sector and analyses the constraints it is facing. Mahangu Grain Trade and Markets Mahangu grain is regularly t raded although in relat ively small quantit ies compared to the overall production, which is mainly home-consumed. In the absence of accurate figures, it is est imated that roughly 10% of production is t raded (Keyler, 1996). Volumes traded vary according to the year’s harvest , but even in years of global deficit some grain is traded. Traded grain is not only surpluses from large communal farmers (even if they wo uld provide the largest part) but also includes small decentralised transact ions at rural village level. The mahangu grain market appears strat ified, with litt le transfer assumed to take place between segments (Presland, 1997). Attempts to quantify volumes and values of the different market segments were made but they remain hazardous in the absence of accurate data. The mahangu market structure can, however, be described as follo ws (adapted from Presland, 1997; Mallet & El Obeid, 1997; Keyler, 1996). Local barter or exchange market At rural household level, grain is bartered for local food stuffs and clothing between neighbours. It is also used to pay church fees in kind. T he exchange unit is very often the lata (around 15 kg) and the price or its barter equivalent are high (from around N$ 20 to N$ 30 per lata). T his kind of exchange can extend to family members in urban areas. Tradit ional mahangu-based fermented beverages are also locally t raded or exchanged. Such transact ions in grain and in processed products involve small quantit ies, but they are likely to be widespread and may represent a significant (although unknown) volume in total. Informal trade market Mahangu grain is sold by producing households for cash to cover basic and urgent needs (school and clinic fees, etc.). Limited quantit ies of mahangu grain are seasonally t raded in the periodic ‘Pension-Day’ markets in rural areas. Local t raders (Cuca shops and retail out lets) and mobile traders buy mahangu grain for cash or in exchange for consumer goods. The terms of exchange do not seem to favour the producers and low cash buying prices have been reported, while re-selling prices are high, making this t rade quite profitable for t raders even if volumes are likely to be limited. Grain t raders indicated a lack of working capital as a limitat ion to this business (Key ler, 1996). Furthermore, mahangu grain and home-processed flour are t raded in latas and cups respect ively on rural and ur ban markets. All reports on this informal t rade suggest that only limited quantit ies are marketed.

Formal trade Different types of operators have been trading mahangu on the local market. Grain traders indicated the period between August and December in the NCRs and bet ween October and December in Kavango as the best t ime for mahangu transact ions. T raders correlated the beginning of the peak selling period with the emptying of rural households’ old mahangu stock. Most commercial mahangu traders intend to store mahangu for less than a year (Keyler, 1996). A mahangu grain market ing survey conducted in the NCRs in July- September 1997, after the bumper harvest of the decade, found literally no mahangu grain t ransactions taking place and commercial buyers un will ing to acquire grain at that t ime, most probably because the demand for grain would have been extremely low after such a good harvest (Mallet & El Obeid, 1997). Most formal t raders are northern-based entrepreneurs, for whom mahangu trading is only a small part of their mult iple business act ivit ies. Farmers’ co-operatives have been playing no, or an insignificant , role in the mahangu trade and there seems to be a very limited number of medium-scale t raders special ising in buying mahangu grain from rural areas. Government has bought mahangu grain locally for its drought relief food distribut ion programmes, as an at tempt to encourage production. T he schemes were implemented by the NDC or private t raders and had an important impact in some years over the last decade, le ss for the quantit ies involved than for the high buying price of mahangu grain, which were reported as having significantly raised price expectat ions from farmers and thus depressed private t rade . The Mahangu Marketing Intelligence Unit , which monitors and facilitates mahangu transactions, reported on a total value of N$ 3 155 000 for mahangu traded during the year May 1999 - April 2000 (Mahangu News). At an average price of N$ 1 100 per tonne, the corresponding traded quantity was just below 3 000 tonnes. Since Keyler’s study in 1992 and 1993, small-scale commercial mahangu millers in the North have emerged as the most consistent and regular buyer of mahangu grain in northern Namibia. Quantit ies processed have, however, remained relat ively low, around 60 tonnes per month in 1999 and around 80 tonnes per month in 2000 (Mahangu News; CRIAA SA-DC, 1999a, 1999c & 2000b). Most of these small-scale commercial millers are large farmers themselves and process their own harvest before buying addit ional grain. It is not clear what quantit ies of mahangu grain are bought on the local market in addit ion to their own harvests. As a newcomer, Namib Mills, the largest industrial processor of maize and wheat in Namibia, has started buying sizeable quantit ies of mahangu grain in Ondangwa and Run du durin g the 2000 marketing season (perhaps over 800 t - 900 t) for their new dry-processed mahangu meal launched in July 2000 and traded as ‘Meme Mahangu’ (the meal contains an undisclosed quantity of maize meal, probably around 20%). T he two buyin g depots were equipped with So uth-African pre-cleaners and mahangu grain was bought at N$ 1 100 per tonne after cleaning (F. Meyer, pers. comm .). T he long-term consumer demand for this new product is not yet established (it has been selling at a rate of around 50 tonnes per month since its launch in July 2000).

Inter- regional and international trade Inter-regional t rade in mahangu is hampered by long distances and high transport costs, especially for grain from Caprivi. A Windhoek-based animal feed company has occasionally bought mahangu from Caprivi using back-load freight , but in irregular and limited quantit ies (36 tonnes at a t ime). Cross-border t rade of mahangu from Angola to the NCRs has been reported but has not been clearly quantified. T he recent lifting of the SACU tariff on mahangu imports may have an impact on this t rade or at least make import stat ist ics more reliable by reducin g the need for shadow or smuggled importat ion. For the past two years or so, South Africa has been a regular importer of Namibian mahangu (around 1 000 t per year) for animal feed and planting as fodder crop. Very recently, according to the MMIU (Mahangu News, Nov. 2000), a large quantity of pearl millet grain, produced under irrigat ion in the Northern Province of South Africa, was prospectively offered for sale to Namibia. Depending on the landed cost of this grain, it could be a major new source of mahangu grain for Namibia. But no transaction has taken place to date and the price asked for this grain seemed too high to be att ractive. No direct exports of mahangu (raw or processed) have been reported to the high-value food speciality market in the northern hemisphere, where a demand exists (C. Lombard, J. Hoffmann, B. Bennett, pers. comm .). Recently, 30 tonnes of mahangu from Caprivi were t raded to the USA through Botswana for an unknown market , assumed to be either for seeds or health food (DoP, MAWRD). Demand also exists for the whole panicles in Europe and North America for decorat ion (sold in flower shops for dry-flower arrangement) and for hanging in birdcages to entertain pet birds; retail prices in these countries appear high (B. Bennett, pers. comm.). E = mc² (Empowerment = mahangu x cash²) ? One important element of mahangu grain t rading relates to prices in the different market segments: • Relat ively high prices can be obtained by producers in the micro-economy at rural

village level, although for small quantit ies. • Bartering and informal t rading does not necessarily ensure a fair price to cash-

strapped rural producers, but t raders in rural areas could ar gue about their high transact ion costs (cash payment, t ransport costs and t ime, storage, etc.) for limited unit quantities traded.

• T he price of mahangu grain in most formal trade transactions is around N$ 1 000 - N$ 1 200 per tonne (or N$ 15 to N$ 18 per lata), which remains largely unat tract ive to significantly st imulate commercial surplus production by small-holding farmers, but is essential to the viability of the commercial processing sector (see section below on commercial mahangu milling).

• Although not directly linked by any price sett ing mechanism, mahangu price formation is dependent on the price of maize in the formal grain market, because maize is the closest subst itute for mahangu. T he maize price in Namibia is itself direct ly linked to the import parity price of South African maize. Mahangu grain has generally been traded in formal markets at a price 10% to 20% higher than white maize grain.

What appears surprising from the monthly price information published by the MMIU is the relative stability of mahangu prices in the ‘formal market’ t ransact ions recorded by the unit throughout the years and between years, with variat ions in price of less than 20% between periods of abundance and t imes of scarcity. It must be said that these records do not take into account the more decentralised and informal sales occurring at farm level and in remote rural areas. But this intra- and inter-annual stability of mahangu prices does not appear to be a st imulat ing factor for private operators, who would struggle to recover the costs incurred in financing and storing stock for re-selling at a later stage. T he ‘market’ seems to behave as if mahangu grain t rade collapses nearly completely when its price tends to rise too high, with maize products taking over. In conclusion, the existing constraints to mahangu grain t rading can be summarised as follows: Low and variable production by smallholder farmers rendering marketed surpluses limited and irregular. Very lo w returns to labour at the current levels of agricultural productivity in NCAs. High price expectat ions from producers related to the low productivity and unreliability of the crop production, but also to a push-up effect from Government mahangu purchase drought relief schemes. Quality of grain t raded is not standardised and prices are not related to grain quality. Information is lacking in the t rade (at both producers’ and traders’ levels) on market demand, prices and quality required by the industry. Producers (who are also the main consumers) are dispersed in rural areas along poor roads at long distances from main towns, making it expensive for t raders to assemble or sell in these places. Few traders in mahangu grain and no integrated network of rural assembly traders and wholesaler special isin g in mahangu trade. High financing and storage cost of mahangu grain for the periods between buying and selling, with an unreliable deman d for mahangu at t ime of scarcity when prices are pushed up. Mahangu grain is more expensive than maize grain, its direct subst itute, throughout the year. Untapped opportunit ies in higher-value mahangu grain (and products) for the local and export market . The Service Milling Sector A service mill (or toll mill) is a business processing the grain of customers against a charge. It is different from commercial milling, where the grain and the processed product belong to the miller and the business consists of processing and sellin g a product for the market . T he commercial milling sector will be looked at in the next sect ion. It must be noted that a number of mills operat ing in northern Namibia are neither service

mills nor commercial mills. T hey are classified as ‘private mills’ in the sense that they mainly operate for the own-use of the owners, who can be communal farmers, inst itut ions such as church missions and schools, or NDC farms. Fieldwork indicated that there is a significant number of these mills. However, they are not reported in the present document. A comprehensive baseline survey in the 6 northern crop-growing regions in July-September 1998 inventoried 87 service mills (CRIAA SA-DC, 1998). This resulted in the establishment of a small millers database at the MMIU to monitor the evolut ion of the small-milling sector. An update survey was carried out in October-November 1999 and identified a total of 128 service mills in northern Namibia (CRIAA SA-DC, 2000b). Anecdotal evidence suggests that this was an under-est imate. Regional distribut ion maps of the mills in Caprivi, Kavango and the North Central Regions are shown in Annex 2. The most important features of the service-milling sector are presented below. Distribution of service mills and evolution in the past years Based on the figures of the two surveys mentioned above, the growth rate of the service-milling sector (with nearly 60 % of all small milling businesses le ss than three years old) has been 56% on average from 1998 to 1999, over a 16 to 17 month period. With a simple arithmetical adjustment, the annual growth rate (over a 12-month period) can be est imated at (at least) 40%. Information collected from various sources in 2000 indicates that this growth trend is continuing. With a total of 116 service mills and an average growth rate of 61 % since the last survey, the NCRs concentrate the bulk of the service milling industry (90 % of service mills and 96 % of new ones). However, the growth rate appears lower in Ohangwena compared to the three other regions: Oshana, Oshikoto and Omusati. With 2 and 0 new mills respect ively, Caprivi and Kavango regions are clearly lagging behind. For each region, the ratio between the number of service mills and the populat ion was calculated as an indicator of the regional ‘density’ of service mills (CRIAA SA-DC, 2000b). The ratios varied from 1/3 000 in Oshana, the most ‘densely serviced’ region, to 1/58 500 in Kavango, the most ‘poorly serviced’ region. Surprisingly, den sit ies in Ohangwena and Caprivi were the same (1/9 000). T he high density of service mills in the Oshana region was mainly due to a large number of service mills in and around the three major towns (Oshakati, Ongwediva and Ondangwa), which have access to good services, part icularly electrical power supply. Out of the 96 non-urban based mills inventoried in the NCRs (there might be more), 39 were concentrated in 17 secondary rural ‘towns’. This is significant in terms of the service milling competit ion developing in these places, which are electrified and are important service centres for rural areas. But it also means that the other rural centres and the rural areas proper are not sufficient ly serviced by mills. It was concluded that service milling, a proximity service, has considerable room for expansion, especially in rural and non-electrified areas (CRIAA SA-DC, 2000b). The two surveys also revealed that service-milling charges had not changed much between 1998 and 1999. In the NCRs, mahangu milling fees generally ranged from N$ 7.50 per lata (N$ 0.50/kg) to N$ 10.00 per lata (N$ 0.70/kg), but could be a s high as N$

12.00 per lata (N$ 0.80/kg). Technology used and technical constraints Almost all service mills are only equipped with hammermills. Most of them have one hammermill operat ing and a few have two. The survey found only three service mills offering dehulling services. Most hammermills (83%) are Drotsky models from South Africa and 50% of all hammermills are the popular M16 models. T he other machines used are imported from neighbouring Southern African countries. A large majority of these hammermills are powered by three-phase electrical motors, 15 kW to 22 kW for the M16 model. Few are driven by petrol or diesel engines, or by tractor PT O. Hammermills can only pulverise the grain, reducing it into flour. For mahangu, the whole grain must be deh ulled before milling ( while it is not necessarily required for maize). Customers of service mills are therefore bringing home-dehulled and fermented mahangu grain, which is most often st ill wet (as reported in the previous chapter). As customers prefer a very fine flour, millers equip their hammermills with fine sieves of 0.8 mm. Hammermilling of wet grain through such fine sieves requires more t ime and energy than dry grain. It is therefore not surprising that service mills are generally equipped with relat ively big machines (such as the M16 with a 15 kW or 22 kW motor) and that the milling throughput is around half (or even less) of the capacity indicated by manufacturers. Optimum capacit ies of the main models of hammermills used in the North were tested by CRIAA SA-DC consultants measuring the actual throughput in real condit ions in a sample of mahangu mills ( CRIAA SA-DC, 1999a). The results have been made available to small millers in the form of an information package, which also includes other useful data and recommendations and was launched in mid-1999 during information and consultat ion workshops for small millers held in the six northern crop-growing regions (CRIAA SA-DC, 1999b). During these consultat ion workshops, the most acute need expressed was for support measures and technical back-up services in the service milling sector (CRIAA SA-DC, 1999d). The baseline survey of 1998 revealed that more than half of the mills were either not working or experienced technical problems, despite more than 40% being less than two years old. In comparison with 1998, the 1999 survey showed a decrease in the number of service mills not running and an improvement of the situat ion after one year. The main reasons are improved installat ion and connection to electrical supply. However, delayed installat ion or electrical supply connect ion still remain major causes of problems. Rural electrificat ion has been progressing in the North Central Regions. It is clear that electricity, when available at the mill, is the cheapest power source and the easiest to handle. Although the nat ional grid is growing, many villages in the North are not yet connected to the grid, especially in Caprivi and Kavango. But some new electrical power lines, especially in the NCRs, have been extended in 220 V single-phase. These are erected to provide basic electrical supply to households and small businesses, but they wil l bring (and already have brought) new problems for small millers planning to run hammermills. First ly, the conventional (three-phased) motors cannot be run with single-phase power supply; and secondly, the electrical power delivered by these lines might not

always be sufficient to run (single-phase) motors of milling machines and is often too weak to start the motors. Examples of maintenance and technical problems identified during the CRIAA SA-DC surveys include: Mill owners buy oversized mills that have motors too large for the available electricity supply and are uneconomical to operate. Suppliers of processing equipment do not provide mill owners and/or operators with sufficient technical and operational training to avoid simple problems. Suppliers are generally not present at the mill to commission the equipment when electricians connect and start the mill. A lack of maintenance skills causes easily preventable problems like hammers not replaced early enough, reversible hammers not turned around when worn out , hammers not replaced as a set (causing imbalance and destructive vibrat ion as a result), wrong pulley alignment and/or belt tension, inadequate or (more often) excessive greasing of bearings, etc. - this leads to exaggerated maintenance and repair costs. Some spare parts are hard to find and/or expensive. There are very few skilled special ist repairers of grain-processing equipment in the regions. No technical manuals or information sheets on operat ions and maintenance are available, which exacerbates the effects of the lack of technical t raining. Mill owners are ignorant about guarantees on new or recondit ioned equipment. Service mill (as well as commercial mill) owners and managers interviewed during these surveys expressed needs for technical training and support , and information on equipment, specificat ions, spares and suppliers (CRIAA SA-DC, 1999d). T he lack of specific technical expert ise regarding grain-processing equipment is part of a general shortage of technical skills in the NCAs resulting from lack of technical exposure and training. A training programme for small millers has been designed (CRIAA SA-DC, 1999e) and is pending implementat ion. Notwithstanding all these technical difficult ies, service milling can be a very profitable business a s long as the turnover is kept above a threshold and technical difficult ies, such as major breakdowns leadin g to stoppages, can be minimised or avoided. This potent ial profitability would explain the continuous growth in the number of service mills in the North. Another factor could be that service milling is a way for wage earners and local business people to re-invest higher disposable income into such a new economic act ivity. For instance, the small millers’ database reveals that some entrepreneurs run a number of service-milling businesses in different parts of the NCRs. Financial viability of service mill ing CRIAA SA-DC consultants carried out an assessment of the financial viability of service mahangu milling in 1999 (CRIAA SA-DC, 1999a). For mahangu service milling, a detailed analysis is provided for the M16A (electric) hammermill (the most common model). It shows that a service milling charge below N$

7.50 per lata (or N$ 0.50 per kg) is an achievable object ive for M16A service mills in the NCRs running at a medium turnover of above 20 latas (300 kg) per day under proper technical and managerial condit ions. However, the business is not viable below an average of 9 latas (135 kg) per day, unless an exorbitant fee above N$ 12.50 per lata (N$ 0.83/kg) is charged. It is also shown that it is technically and financially feasible for higher turnover service mills, over 42 latas (630 kg) per day, to charge much less, around N$ 4.50 per lata (N$ 0.30 /kg). Other models of hammermills were not all suitable for service milling in northern Namibia, due to technical, purchase costs and other financial viability reasons. As the demand for service milling is very likely to continue to expand, further research into specific aspects of mahangu service milling was recommended by CRIAA SA-DC consultants, including an assessment of technical options and financial condit ions of service milling in non-electrified areas usin g alternat ive power sources (CRIAA SA-DC, 1999a & 1999c). Economic importance of service mill ing The development of the service-milling sector has mainly been the results of private sector init iat ives with very limited direct support from Government and development aid. However, this development was made possible by a conducive environment provided by Government encouraging the private sector, developing basic infrastructure, such as the electrificat ion in Northern Namibia, and facilitat ing access to credit for small entrepreneurs. Although employment figures in the service-milling sector may look modest (direct employment has been est imated at around 350 at end of 1999), its contribution to the Northern economy is far from marginal. Furthermore, by their numbers service mills provide the ‘market size’ that makes possible the provision of private sector services to the small-scale milling sector, commercial mills included. It is not clear at present if service dehulling has a good chance to develop in Northern Namibia. As explained in the previous chapter, the majority of mahangu consumers prefer the fermented flour, which implies, in tradit ional food processing, the soaking of the dehulled grain before milling. Customers of service mills wo uld have to bring their grain for dehulling, then take it back home for soaking, and bring the fermented grain back to the mill (a second t ime) to get the flour. Unless customers drop their food preparation customs (which seems unlikely and nutrit ionally detrimental) or entrust millers to soak their grain at the mill (which might not be logist ically easy and may not be accepted by customers), the development of service dehulling will mean having customers going to the mill twice, once for dehulling and later for milling. In conclusion, service mills have the most pressing technical problems and difficulties in business planning and management, they are by far the highest number of processing units and their number is gro wing regularly with new installat ions. It was est imated that the actual throughputs of service mills do not exceed 25% of their opt imum technical capacity. T his has detrimental implicat ions for their viability. Improving the efficiency of service mills will facilitate a reduction in service charges in the Northern Communal Areas. This will not only benefit customers, but increase demand and thus improve the turnover of the mills (which seldom run at capacity). Reliable and affordable service

milling saves rural women the drudgery of manually pounding grain for the daily meals and frees them to pursue more productive activit ies, direct ly benefiting farm family well-being. The Commercial Mil ling Sector CRIAA SA-DC consultants identified 13 small-scale commercial mills operat ing in the NCAs at the end of 1999. Six of them were only milling maize, three were mixed units processing both maize and mahangu (dry process), and four were only milling mahangu (wet process), two of which were less than one year old. Maize milling, carried out with the same type of machinery as mahangu milling but also with rollermills, is not discussed further in this report. It must be noted, however, that the installed capacity of small commercial maize mills in the NCAs exceeds the present demand and that at least three other milling companies further south also supply the Northern market with maize meal. Most of the constraints experienced by service millers are also faced by commercial mahangu millers and they will not be repeated. In addit ion, commercial mahangu processors experience specific difficult ies, which are presented below. Regional distribution of small commercial mahangu mills and installed capacity The mahangu commercial mills are few in number and can therefore be named. At the end of 1999, there were two in Kavango (Kamalanga Milling at Katjinakat ji and I.J. Mills at Nkurenkuru), one in Oshikoto (ABC Mill at Onyaanya), and two in Omusati (Okavu Mill at Outapi and M.N. Mill in Uukwaludhi). T wo new commercial mahangu mills started operat ing in 2000, T S Propert ies near Oshikango and Sun Moonlight at Ohangwena. With the type and number of processing equipment used in these commercial mills, the installed capacity was 88 tonnes per month in 1999, or around 1 000 tonnes per annum. The actual production in 1999 was estimated between 45 tonnes and 53 tonnes per month (540 t to 640 t per annum). The mahangu meal installed capacity and current production remained marginal compared to maize meal (CRIAA SA-DC, 2000b). Factory gate prices in 1999 for the ‘wet process’ mahangu meal were between N$ 2.40 per kg (large bags) and N$ 3.20 per kg (small bags). These prices remain much higher than for maize meal (CRIAA SA-DC, 1999a & 2000b), as shown in Table 12 below. Table 12. Comparative Prices (N$ / tonne) of Maize Meal in North Central Namibia (1999) Whole hamm ermilled

Dehulled and hamm ermilled

Industri al unsi fted *

Small-scale rollermilled

Industri al sift ed *

Industri al speci al si fted *

1 400 1 600 1 400 1 600 1 800 2 000 * Wholesale price, othe rwise factory price when produce d in North Central Namibia Source: Trade survey (CRIAA SA-DC, 1999a). The different commercial methods of processing mahangu There are current ly two mahangu processing methods followed by commercial mahangu millers in (northern) Namibia:

Dry processing, in which mahangu grain is pre-cleaned, mechanically dehulled with an abrasive deh uller, hammermilled and bagged straight away. At no stage is the mahangu humidified. The wet process, in which mahangu is soaked and (slight ly) fermented after dehulling and before milling. At some stage(s) in the process, the product must be dried to avoid the bagged en d-product rott ing. This is a much more complex way of processing than the dry method, but the product has a higher value on the market. T he wet processing was init iated, on a small-scale commercial basis, by Mr and Mrs Kamalanga, and is st ill bein g developed and improved (Special Annex in CRIAA SA-DC, 1999a). T he consumer market is very sensit ive to the quality of the fermented mahangu meal (colour, taste, odour, texture, etc.); there could be a s many different meals as there are processors, each following its own recipe. Technology used and technical constraints on commercial mahangu processing Processing pearl millet is more difficult than milling maize (and sorghum): mahangu grains are smaller in size, they must be dehulled at a lower extract ion rate and the milling yield is lo wer. The average composit ion of pearl millet grain, as compared to sorghum and maize, is provided in Table 13 below. Furthermore, mahangu has a higher oil content than maize (and sorghum). Mahangu bran, which contains part of the oil-rich germ, easily st icks to dehullers parts, part icularly to the bran extract ion system and can block it, as can be witnessed on most mahangu deh ullers operat ing in northern Namibia. Table 13. Average composition of cereal grains (percenta ge of total dry weight) Pearl millet Sorghum Mai ze Whole grain 100 100 100 Endosperm 75 82 82 Germ 16 10 13 Bran 9 8 5

Adapted f rom Abrasive-Disk De hullers in Africa (1989) and Small-Scale Milling (1994) Mechanical dehullin g remains a bot t leneck in increasing the output of commercial mahangu mills. T he different models of (abrasive) dehullers used in the NCAs were tested by CRIAA SA-DC consultants by measuring the optimum throughput in real condit ions at a sample of mills (CRIAA SA-DC, 1999a). T he conclusions were that although one model was performing better than the others, none were adapted to and sat isfactorily processing mahangu. Research into and development (or adaptat ion) of appropriate mahangu dehullers was therefore recommended as priorit ies. CIRAD consultants noted that due to the small size of mahangu grain no perfect balance had ever been found with small-scale dehullers between polishing, cleaning and high throughput . T hey further explained that developing a new dehuller is a complex task, due to at least ten technical design parameters to be taken into account , and requires a rigorous technological approach. R&D costs may only be recovered if the number of machines manufactured and disseminated is sufficient . T est ing, adapting and improving available equipment might be an easier path, but st ill requires professional technological skills. The CIRAD consultants further noted that in commercial milling, two or three medium-sized dehuller s running in parallel should be preferable to one large-sized

dehuller, as they would give more flexibility and wo uld al low service dehulling aside (Goli & Ndiaye, 2000). Other major difficult ies experienced by ‘wet’ mahangu millers are the handling of soaked grain and the drying of the flour, for which no appropriate mechanised option (or at least improved handling method) is presently available. As mentioned earlier in this report , commercial millers have often complained about the qua lity of mahangu grain purchased from farmers and the contamination of grain with sand, other foreign matter and insect infestat ion. T o provide a regulatory framework, the nat ional standards for mahangu grain were recently promulgated in the Government Gazet te (Republic of Namibia, 2000). Among all the commercial mahangu mills, only two mills were using grain pre-cleaners in 1999, while the rest experienced excessive equipment wear caused by abrasive san d and dust. The electrical vibrat ing-sieve models used were, however, not efficient ly cleaning the mahangu grain due to poor design (an addit ional sieve lacking) and required addit ional labour to manually remove foreign matter of larger size. Financial viability of commercial mahangu milling CRIAA SA-DC consultants carried out an assessment of the financial viability of commercial mahangu milling in 1999 (CRIAA SA-DC, 1999a). It was concluded that the dehull ing extract ion rate had some influence on the cost of the end product , but much less than the cost of raw material procurement. Over 60 % of the production cost of mahangu flour is represented by the costs of the raw material, and its financing and storing. The price of mahangu and the cleanness of the grain are the main cost factors of commercial mahangu milling (dry and wet process). According to this financial appraisal, at a raw material price of N$ 1 000 per tonne for uncleaned mahangu grain, the total production costs (bran income deducted) of dry processed mahangu flour cannot be lower than N$ 1 950 per tonne. If a reasonable enterprise profit (before tax) of 20% is added, the ex-factory price of mahangu flour cannot be lower than N$ 2 300 per tonne. With maize meal wholesale prices generally ranging from N$ 1 400-1 600 per tonne for unsifted meal to N$ 1 900-2 000 per tonne for sifted meal, the commercial production of mahangu flour, through the dry processing method, at a parity price with maize meal , is not viable. With mechanised dehullin g, the financial analysis showed that it is difficult, if not unfeasible, to produce commercial wet processed mahangu meal be low a break-even cost (bran income deduction included) of N$ 2 000 - 2 100 per tonne, and consequently to sell it , in bulk (50 kg bags) at the factory gate, at a price below N$ 2 400 per tonne, if a profit (before tax) of around 20 % is expected (all other costs being covered). The same conclusion was drawn for the wet processing method, for which the production of mahangu meal at a parity price with maize meal is also not viable. Further research and development needed The technical and financial appraisal of wet mahangu processing operat ions shows room for improvement from a productivity and financial point of view. CRIAA SA-DC consultants recommended further research and support into improving the commercial mahangu wet processing technology. In view of promoting high-value products from

mahangu processing, it was also recommended to extend the research to other processing methods and products other than flour or meal. A preliminary list of potent ially interest ing products was e stablished, but it was noted that consumer acceptance for these products would have to be assessed at an init ial stage, before expensive research and development efforts are invested into these new product lines (CRIAA SA-DC, 1999a). The slight ly fermented (acid/lact ic fermentat ion) mahangu meal remains a ‘luxury’ product for a specific segment of the market. Soaking and fermentation are, however, reported in the literature to improve the digestibility and the nutrit ional content of cereals, which gives higher value to such products. CRIAA SA-DC consultants recommended that further research be conducted into soaking/fermenting of dehulled mahangu grain, and into the nutrit ional composit ion of wet-processed flour compared to dry-processed flour (CRIAA SA-DC, 1999a). CIRAD consultants concurred with the above recommendations and noted that fermentat ion processes have been studied in most West African countries. Most often, ‘hetero-fermentat ive’ fermentat ion takes place, the main strain being Lactobacillus plantarum . Fermentat ion could be accelerated and standardised using a starter (selected cultures or an inoculum from a previous batch) and maintaining the temperature above 25°C (Goli & Ndiaye, 2000). The consultants from CIRAD also stressed the importance of ensuring a regular qua lity of processed products, part icularly the wet processed flour, through specificat ions and standards, and promoting good manufacturing pract ices. A nutritional analysis of Kamalanga’s flour was organised by the Namibian Agronomic Board (see Table 14 below). Unfortunately, there is no such analysis for a dry process mahangu meal as a basis for comparison. Table 14. Analysis of a mahangu flour sample from Kamalanga Mill (May 1998) Component (method) Units Results Moisture g / 100 g 13.1 Ash g / 100 g 1.30 Diet ary fib re (t otal) g / 100 g 6.48 Carbohydrat e (by di fference) g / 100 g 63.3 Fat (acid hydrolysis) g / 100 g 5.00 Protein (N x 6.25) g / 100 g 10.8 Energy (cal cul ated ) kJ / 100 g 1 381 Calcium mg / 100 g 163 Phosphorus mg / 100 g 196 Potassium mg / 100 g 177 Iron mg / 100 g 6.52 Magnesium mg / 100 g 63.8 Niacin (Vit. B3/PP) mg / 100 g 1.15 Riboflavin (Vit. B2) mg / 100 g 0.056 Thiamin (Vit. B1) mg / 100 g 0.22 Source: Di vision of Food Science and Tec hnology, CSIR (Pretoria), courtesy of NAB . Utilisation of commercial mahangu mill products

Mahangu products from small-scale commercial mills in Northern Namibia have so far been essentially consumed in the form of meal for home consumption, as is the new mahangu-maize mixed meal of Namib Mills. Namib Mills’ meal is on the shelves of a good number of supermarkets and medium-sized retail shops in Windhoek, in the North and in major urban mahangu consumption centres, alongside the range of maize meals usually distributed in Namibia. Mahangu meals from small-scale commercial mills are also available in northern supermarkets and retail shops, as well as in Windhoek (part icularly in the medium-sized supermarkets of Katutura) and in major urban centres. Northern commercial millers seem to have specific market niches, since only one brand is generally found in a part icular retail out let (often alongside the Namib Mills mahangu mix). As was pointed out above, small commercial mahangu mills currently produce and market at most 100 tonnes of flour per month and the new Namib Mills mix only sells at around 50 tonnes per month. The present domestic consumption of commercial mahangu meal can therefore be est imated at around 150 tonnes per month, at most . Consumption figures may change in future. It could also increase further in the present market ing year as consumers tend to purchase milled products at a higher rate towards the end of the calendar year and in the first semester of the new year before the new harvest . It is presently not clear, at the current price of mahangu meal and with the current quality of products offered on the market, whether the domestic market for mahangu flour has been saturated or whether there is room for expansion. Certainly, in-depth consumer surveys wo uld provide a bet ter understanding of this issue. It is known that commercial mahangu millers presently face market ing difficult ies and have to restrict their production to immediately marketable levels, because of the relat ively short shelf life of their products. The institut ional demand (schools, hospitals, prisons, army, mines, etc.) has not yet been significantly tapped by commercial mahangu millers, who have faced various difficult ies in accessing these potent ially sizeable markets. One major difficulty has been that these inst itut ions procure foodstuffs on tender through catering companies, which bid for and ult imately procure the cheapest products. Maize meal always had a price advantage over mahangu meal. Efforts by Government to give preference to mahangu meal on Government inst itut ion tenders have not yet brought results. It has, however, been set as a priority intervention of Government to st imulate the commercialisation of the crop, encourage the development of commercial processing and just ify invest ing into improved processing technology. T he major constrain has been the higher cost of procuring mahangu by caterers, which would ultimately be passed on to Government. Compulsory inclusion of mahangu products in Government procurement tenders wo uld also face two types of difficult ies. First ly, mahangu is not preferred by all the Namibian populat ion and is often negatively perceived as a backwar d tradit ional food (of the rural North). Secondly, there is limited capacity and human resources in Government to actually control the composit ion of rat ions supplied by caterers. On a more posit ive note, it must be mentioned that a major mining company and Kamalanga Mill have been discussin g the supply of mahangu meal to mine canteens on a regular basis (around 20 tonnes per month). New products made of mahangu are still at an infant stage (bread with 20% mahangu

flour, biscuits or pasta with mahangu flour), with the exception of the ‘Oshikundu’ beverage of Namibia Breweries, which has so far not sold at the level expected by the company. T he Tunweni Brewery in Tsumeb has been manufacturing a fermented drink made with mahangu meal, sugar, lact ic acid and water, pasteurised and packaged in 500 ml cardboard packs, and traded as ‘Oshikundu’. It is current ly only using 1.5 tonnes per month of wet processed mahangu flo ur bought from Kamalanga mill in Kavango (J. Dammert in MAWRD, 2000). It could be argued that as long as the potent ial domestic demand for mahangu flour has not been sat isfied, there is no need to embark on developing new mahangu products. T his considerat ion is strengthened by the fact that the supply of mahangu raw material at a commercially viable cost is presently a limit ing factor to the industry. But it could also be argued that higher-value products could bet ter accommodate supply price variat ions and could also respond bet ter to the changing demand patterns of consumers, especially in urban areas. CHAPTER 5 MAHANGU PROMOTION POLICY HIGHLIGHT S AND DISCUSSION From the Workshop to Develop a Strategy for the Multiple Use of Mahangu and Sorghum in Namibia referred to in Chapter 1, a ‘Strategy and Action Plan’ was elaborated by MAWRD and stakeholders (MA WRD, 2001). A large port ion of this strategy deals with post-harvest issues, which will not be elaborated in detail below as it was in the main text of this document. Institutional arrangements, progress review and re -planning This research report has tried to demonstrate from a post-harvest angle that constrains to mahangu promotion are numerous and largely interlinked. It is typically a complex problem, which requires a mult i-pronged approach. As room for manoeuvre is limited at all the identified constraint levels, the strategy must be holistic and the act ion plan targeted at lift ing sets of bott lenecks. The strategy therefore needs to be not only relevant and mult i-disciplinary, but also prioritised (as opposed to loosely mult i-direct ional), while the implementat ion of the act ion plan needs to be co-ordinated and from t ime to time evaluated. T he Action Plan recommended the formation of a Mahangu (and Sorgh um) T ask T eam (MSTT ) lead by MAWRD and including governmental inst itut ions and non-governmental organisat ions. Primary agricultural production Although it was not the focus of this research, low and variable primary production remains a fundamental problem hampering the commercialisat ion of mahangu. T here is certainly room for increasing agricultural productivity (not only the yield per hectare but also the return to labour) through improved cult ivars, mechanisat ion (animal t ract ion and tractor services), increased manure rest itut ion to cult ivated soils, etc. T here are also opportunit ies to spread risks and diversify the ‘cropping basket’ in such adverse agro-ecological condit ions as in Northern Namibia. T hese const itute the primary mandate of

the agricultural extension service fed by recommendations from agricultural research and farming system research units. It must be stressed that only marginal productivity gains may be possible because of agro-ecological limitations, but such gains might st ill be significant overall if spread among a large port ion of communal farming households. In the very densely populated areas of Ohangwena and Oshana regions, where rural populat ion densit ies can reach 100 inhabitants per square kilometre, absolute limits on the availability of land for cult ivat ion (relat ive to cereal consumption in these areas) might further restrict the commercial impacts of increasing yields and/or returns to labour. The mahangu promotion strategy advocates as one of its component encouraging increased market-orientated production of mahangu (and sorghum) on land where these crops have not been tradit ionally grown. Knowing the agro-ecological potent ial of Namibia, these areas can only be the ‘commercial farming’ areas of the maize triangle, where production of maize and other food crops (rain-fed and irrigated) has been constantly shrinking with the gradual liberalisat ion of Namibian trade in agronomic commodit ies. The strategy is not clear about whether this new commercial mahangu production should be encouraged amongst the exist ing better-off ‘commercial farmers’ or should be the basis of the set t lement of small-scale farmers orientated towards commercial mahangu (and other cash crops) production. It is not the purpose of this research paper to enter into this debate, which concerns another set of policies, but it must be highlighted as it is relevant to the present mahangu strategy. Post-harvest processing and grain storage Going back to post-harvest issues, an important component of the strategy is the reduction of post-harvest losses by improving tradit ional drying, threshing and storage methods. T his is certainly a priority area because problems have been clearly identified, the impact can be widespread amongst producers, and it has the potent ial to increase the effect ive harvest (and its quality) available for home-consumption and market ing. The strategy stressed the lack of applied research into post-harvest issues and the subsequent lack of extension recommendations as major impediments. It is expected that the creat ion of a post-harvest research focus group or sub-division with MAWRD as recommended durin g the Mokuti Workshop, as well as the integrat ion of post-harvest issues in FSRE and extension services work, wil l address these shortcomings. Mahangu grain marketing and processing As a result of low primary production, very limited quantit ies of mahangu are available nat ionally for t rading, which in turn retards the development of market ing systems and severely restricts the options for secondary processing and value-adding. Combined with the fact that pearl millet is not readily available on the internat ional market and cannot be imported at a price competit ive with maize, the limited trade in mahangu grain renders processing businesses very vulnerable to failure due to an interrupted supply of raw material. The grain marketed is often of poor and variable quality, result ing in unacceptably high processing losses. These constraints are reflected in the nature of the recent growth that has occurred in the processing sector: it has primarily been focused on service milling rather than commercial milling.

On the other side of this vicious circle, the underdeveloped commercial mahangu processing industry has contributed significantly to the absence of a market incentive to increase primary production and trading. A major posit ive considerat ion is the very strong consumer preference, and thus demand, for mahangu products, especially the slight ly fermented mahangu meal produced by wet processing. Low production and high demand theoretically t ranslate into higher prices, a trend confirmed by the premium price realised by such mahangu meal in urban markets. The main strategies that have been suggested as ways to facilitate and support increased trade in and processing of mahangu include: to support mahangu grain market ing through bet ter market information to sellers and buyers (e.g. st rengthening of MMIU), improved off-farm storage and trading facilities (assembly depots in surplus producin g areas, wholesale rural markets), encouragement to local processing, support to mahangu grain t raders and market ing co-operatives (loans, ...). to invest igate increased imports of mahangu from southern Angola in years of nat ional deficit and encourage the inter-annual stockpiling of mahangu grain to ensure supply in drought years. to require that a minimum percentage of mahangu-based products be included in Government and other inst itut ional food-supply contracts. to develop technical and business t raining for the small-scale milling sector, to provide improved technological opt ions for processing equipment. Value-adding and product development Only very limited quantit ies of Namibian mahangu (and sorghum) are current ly used to make value-added products for formal markets (if mahangu meal is excluded, the quantit ies are insignificant). It has been suggested that the development of innovative, high-value products will st imulate increased mahangu (and sorghum) production, market ing and pre-processing, by providing a constant and lucrat ive market for grain, meal and malt . Since it is not possible to produce mahangu cheaper than imported maize, there is a need to develop higher-value mahangu products that will fetch a premium price in niche markets. The main strategies that have been recommended in this regard are: product development. market promotion (generic and brand-specific). incentives schemes to manufactures. In pursuing the product-development approach to mahangu (and sorghum) promotion it is important to guard against disrupting the numerous informal-sector micro-enterprises current ly producing tradit ional beverages and foods from these grains. T his can be done by ensuring that the new products are t ruly innovative and compete in different market segments. Developing new products and successfully market ing them require that a set of

condit ions be fulfilled. T hese include formulat ing good quality products that meet the expectations of consumers (in-depth understanding of consumer perceptions is essential), and targeted promotion campaigns at the launch. Conclusion Improving mahangu post-harvest systems is a necessary condit ion for the growth and expansion of the mahangu industry in Namibia, but it is not in itself a sufficient condit ion. Better post-harvest pract ices and technology can increase the effect ive nat ional harvest and the efficiency of processing, but ult imately the sustainable gro wth of the secondary and tert iary mahangu sectors will need to be supported by increased primary production and/or more reliable and cheaper imports of mahangu grain, especially in drought years. However, many of the constraints on mahangu post-harvest systems have been identified and addressing these constraints is there fore a priority area for immediate intervention. * * * BIBLIOGRAPHY PROJECT REPORTS TO DATE • CRIAA SA-DC (1998) “Baseline Study on T echnical Support to the Small Scale

Mahangu and other Grain Processing and Marketing Industry in the Northern Communal Areas”, Final Consultancy Report , GAMOND R. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, October

• CRIAA SA-DC (1999a) “Technical Support to the Small Scale Mahangu an d other Grain Processing and Marketing Industry in the Northern Communal Areas, Final Report Phase II-A”, GAMOND R. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, April

+ [”Annex 9: Mahangu Wet Processing Based on Kamalanga Commercial Milling Operations” submitted separately to protect commercial confidentiality] • CRIAA SA-DC (1999b) “Small Sca le Millers Consultat ive Meetings Information

Package”, prepared for NAB and Directorate of Planning (MAWRD), Windhoek, June

• CRIAA SA-DC (1999c) “Technical Support to the Small Scale Mahangu an d other Grain Processing and Marketing Industry in the Northern Communal Areas, Addendum to the Consultancy Report of Phase II-A: Quantification of the Recommendations”, GAMOND R. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek July

• CRIAA SA-DC (1999d) “Technical Support to the Small Scale Mahangu and other Grain Processing and Marketing Industry in the Northern Communal Areas, Final Report Phase II-B”, GAMOND R. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, July

• CRIAA SA-DC (1999e) “Training Funding Proposal For Submission to MAWRD - Directorate of Planning, Final Report Phase II-C (Component No 1)”, Du PLESSI S P. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek November

• CRIAA SA-DC (2000a) “ Implications of Extended T erms of Reference for the MMIU, Final Report Phase II-C (Component No 2)”, Du PLESSI S P. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, January

• CRIAA SA-DC (2000b) “1999 Update of the Small Millers Database, Final Report Phase II-C (Component No 3)”, GAMOND R., Du PLE SSI S P. & MALLET M. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, January

• CRIAA SA-DC (2000c) “Mahangu Pre-cleaner and Grader Design and Technical T est Report, Final Report Phase II-C (Component No 5)”, GAMOND R. (CRIAA SA-DC) for NAB and Directorate of Planning (MAWRD), Windhoek, August

DOCUMENT S CONSULT ED ON PEARL MILLET AND POST HARVEST • BALOGUN P. & TRIPP R. (1997) “An Analysis of Namibia’s Present and Future

Requirements for Seeds and Options for Supplying those Needs”, ODI, London, (Draft) March

• BASSEY M.W. & SCHMIDT O.G. (1989) “Abrasive-Disk Dehullers in Africa, from Research to Dissemination”, IDRC, Ottawa, Canada

• BEZANÇON G., RENNO J-F., ANAND KUM AR K. (1997) “Le mil” in L’amélioration des plantes tropicales, CIRAD & ORST OM, Toulouse, France

• BLENCH R.M. (1997) “Pearl Millet in Developing Countries”, Web page, ODI, London, November

• BROUT IN C. & SOKONA K. (1999) “ Innovations pour la promotion des céréales locales: Reconquérir les marchés ur bains”, GRET & ENDA-GRAF, Dakar, Sénégal, November

• COULT ER J. (1994) “Liberalizat ion of Cereals Market ing in Sub- Saharan Africa: Lessons from Experience”, NRI Marketing Series 9, NRI, Chatham, UK

• CRIAA (1996) “Crop Processing Services of Likwama Farmers’ Co-operat ive Union (LFCU): T hreshing Service”, Implementat ion Report to the EC Food Aid Counterpart Funds, Windhoek, March

• DENDY D.A.V. (1995) “A Study of Millet (Mahangu) Processing in Namibia: T he Way Forward”, prepared for MAWRD, Grains After Harvest, Oxfordshire, UK, April

• DENDY D.A.V. (1993) “Report on a Visit to Namibia to Assist in the Improvement of Pearl Millet Processing and Products at the NDC, Musese, Kavango”, NRI, Chatham, UK

• DENDY D.A.V. ed. (1977) “Proceedings of a Symposium on Sorghum and Millets for Human Food”, held at the 9th Congress of the Internat ional Association for Cereal Chemistry in Vienna, 11-12th May 1976, Tropical Products Inst itute, London

• ENDA-Zimbabwe (undated, 1994?) “Prospectus: Small Grain and Maize Mill ing”, Harare

• ENDA-Zimbabwe (undated, 1994?) “Dehuller Owner / Operator’s Manual”, Harare • ENDA-Zimbabwe (1991) “Mission Report to the Republic of Namibia, 14-21 April

1991”, GORE C. & CHAGUM A A., Harare • ENDA-Zimbabwe (1988) “T est Market ing of Small Grain Meal Products in

Zimbabwe”, presented to IDRC (Canada), Harare, January • ERIOLA M. & BRADLEY J. (1990) “Kavango: the Sambiyu T ribe, the Way of Life

of the Mupapama River T errace Community”, FINBATT-UNT AG, Rundu, Namibia • FAO & INPhO (1995) “Le sorgho et les mils dans la nutrit ion humaine”, Collect ion

FAO: Alimentat ion et nutrit ion no 27, FAO, Roma • FAO/ICRI SAT (1996) “T he World Sorghum and Millet Economies: Facts, T rends

and Outlook”, FAO, Rome and ICRI SAT, Patancheru, India • FISCHE R-BUDE R V. (1999) “Report on a Suitability Assessment of Six Sites in

Northern Namibia for the Establishment of Bulk Grain Handlin g and Storage Facilit ies”, prepared by Stewart Scott Namibia for NAB, Windhoek, July [and addit ional report of September 1999]

• FOWLER M. (1997) “Food Security or Food Self- Sufficiency for Namibia? The Background an d a Review of the Economic Policy Implicat ions”, Discussion Paper no 1, Division of Agricultural Planning, Directorate of Planning, MAWRD, Windhoek, June

• GELDENHUYS F.D. & al. (1997) “Hammer Mill Manual and Test Report”, Inst itute for Agricultural Engineering (ILI/IAE), Agricultural Research Council (LNR/ARC), Si lverton, South Africa

• GOLI T . & NDIAYE J-L. (2000) “Mahangu, Sorgh um and Indigenous Fr uits Processing and Promotion in Namibia”, Mission to Namibia (23/10-3/11/2000), consultancy report for MAWRD and French Co-operat ion, CIRAD-AMIS, Montpellier, France, December

• GRET (undated) “Moulin à Marteaux, environnement, indicat ions techniques”, document d’ingénierie, Paris

• HAY R., DENDY D., MART IN A., WHIT COMBE J. (1991) “Report on a Visit to Namibia to Carry Out a Feasibility Study of the Production and Marketing of Pearl Millet and its Products”, NRI, Chatham, UK

• HELMST ETT ER D. (1997) “ Involving the Private Sector in Improving Farming Systems: the NNRDP Experience”, Northern Namibia Rural Development Project (NNRDP), Ongwediva, Namibia, January

• HERRM ANN G. & NGHIPANDULWA P. (1995) “Needs Assessment Mission: T raining Programme for Women Entrepreneurs in the Food Processing Sector, Namibia”, final report , UNIDO, Windhoek, March

• HINDMARSH P. (2000) “Impact of Store Design and Management on Reduc ing On-farm Losses in Pearl Millet”, report on field t rials to MAWRD, NRI, Chatham, UK

• ICRISAT (1996) “Second National Pearl Millet Workshop: Proceedings of a Workshop, 7-8 Nov. 1994, Windhoek, Namibia”, SADC/ICRI SAT Sorghum and Millet Improvement Program, Bulawayo, Zimbabwe

• INTERMEDIATE T ECHNOLOGY (1994) “Cereal Processing”, Food Cycle T echnology Source Books, IT Publicat ions & UNIFEM, London

• IPINGE S.A. (2000) “Overview of the National Sorghum and Pearl Millet

Improvement Programme” paper presented at the Workshop to Develop a Strategy for the Multiple Use of Mahangu (Millet) and Sorghum in Namibia, 27-28 June 2000, Mokuti Lodge, Namibia

• JAGO N.D. ed. (1993) “Millet Crop-Loss Assessment Methods”, NRI Bullet in 62, Chatham, UK

• JONSSON L-O., DENDY D., WELLINGS K., BOKALDE RS V. (1994) “Small-Sca le Milling: A Guide for Development Workers”, IT & Stockholm Environment Inst itute, London

• KALISKY J. (1992) “Equipment for Post-Production Operat ions of Cereals and Pulses”, AGROT EC, UNDP/OPS Regional programme, Harare, September

• KANDANDO R.J. & NGWIRA T.N. (1999) “Consumption of Pearl Millet (Mahangu) Products in Namibia”, consultancy report for FAO Namibia, UNAM, Faculty of Agriculture and Natural Reso urces, Windhoek, (draft) August

• KANT É A. (2000) “Strategies for Developing Cereals’ Grain Market ing and Experience in Diversificat ion the Use of Sorghum and Millet in Mali” presented at the Workshop to Develop a Strategy for the Multiple Use of Mahangu (Millet) and Sorghum in Namibia, 27-28 June 2000, Mokuti Lodge, Namibia

• KEYLER S.K. (1996) “Economics of the Namibian Millet Subsector”, PhD dissertat ion, Department of Agricultural Economics, Michigan State University, East Lansing, USA

• KFSR/E (1998) “Annual Report 1997”, Kavango Farming System Research and Extension, Mashare, Namibia

• KROLL T . (1991) “Prospects on Introduction of New Appropriate Post Harvest Millet Processing Technologies in Northern Namibia”, UNICEF, Windhoek

• LABET OULLE L. (2000) “Nutritional Survey of Rural Communit ies in the North Central Regions of Namibia”, Research Report to French Mission for Co-operat ion and DEE S - North Central Division (MAWRD), CRIAA SA-DC, Windhoek, March

• LARSEN A. ed. (undated) “Crop Pests in Northern Namibia: Pearl Millet, Sorghum, Maize, Cowpea and Cucurbits”, Training Manual no 1, Farming Systems Research & Extension Unit (North Central Division), MAWRD, T sumeb, Namibia

• LOW A. (1995) “Linkages between Maize Market ing Margins, Household Food Security, and Mahangu Production” in Second National Pearl Millet Workshop: Proceedings of a Workshop, 7-8 Nov. 1994, Windhoek, Namibia”, SADC/ICRISAT Sorgh um and Millet Improvement Program, Bulawayo, Zimbabwe

• MALLET M. & EL OBEID S. (1997) “Marketing of Mahangu by the Northern Namibia Farmers’ Co-operat ive Ltd (NNFC): Feasibility Study”, consultancy report for NNFC & Division of Co-operat ive Development (MAWRD), CRIAA SA-DC, Windhoek, November

• MASDAR ZAMBIA Ltd (1993) “Northern Regions Study”, Northern Regions Development Programme (NRDP), interim report (Sep.) & draft final report (Nov.), MAWRD / ADF, Windhoek

• MATANYAIRE C.M. (undated, 1998?) “Understanding the Pearl Millet Production Environment in the Mixed Farming Areas of Northern Namibia”, Rural Development Support Programme (RDSP), Oshakati, Namibia

• MATANYAIRE C.M. (1998) “Sustainability of Pearl Millet (Pennisetum glaucum ) productivity in Northern Namibia: Current Situation and Challenges”, in So uth

African Journal of Science, vol. 94: 157-166, South Africa, April • MATANYAIRE C.M. (1996) “Pear Millet Production Systems in the Communal

Areas of Northern Namibia: Priority Research Foci Arising from a Diagnost ic Survey”, in Drought-tolerant Crops for Southern Africa : proceedings of the SADC/ICRISAT Regional Sorghum and Pearl Millet Workshop, 25-29 July 1994, Gaberone, Botswana (43-58), ICRISAT , Patancheru, India

• MAWRD (2001) “Promoting Mahangu and Sorghum in Namibia”, prepared by CRIAA SA-DC (P. du Plessis ed.) for NAB and Directorate of Agricultural Research & Training (MAWRD), Windhoek, January

• MAWRD (2000) “Proceedings of a Workshop to Develop a Strategy for the Mult iple Use of Mahangu (Millet) and Sorgh um in Namibia, Mokuti Lodge, 27-28 June 2000”, prepared by CRIAA SA-DC for NAB and MAWRD, Windhoek, August

• MAWRD (1999a) “Farm-Household Economics Survey of the Caprivi Region Namibia: Analysis Report 1, Sep. 1996 - June 1997”, Directorate of Planning, Windhoek, March

• MAWRD (1999b) “Rural Livelihood Diversity in Kavango: Quantification of Alternat ive Sources of Farm-Household Consumption and Production”, discussion paper No 3, Directorate of Planning, Windhoek, May

• MAWRD (1999c) “Proceedings of the National Annual Agricultural Research Report ing Conference of DART , held on 6-10 Sep. 1999 at Swakopmund”, Directorate of Agricultural Research and Training (DART ), MAWRD, Windhoek

• MAWRD (1998) “Pearl Millet Improvements: Plant Material Descript ion”, Spotlight on Agriculture 7, DART , Windhoek, April

• MAWRD (1997) “Farm Management Survey of the Okavango Region: Analysis Report 1, Nov. 1995 - June 1996”, Directorate of Planning, Windhoek, March

• MAWRD (1995) “National Agricultural Policy”, GRN, Windhoek • MENDEL SOHN J., EL OBEID S. & ROBE RT S C. (2000) “A Profile of North-

Central Namibia”, Environment Profiles Project , Directorate of Environmental Affairs, Ministry of Environment and T ourism, Gamsberg Macmillan, Windhoek

• MENDEL SOHN J. & ROBE RT S C. (1997) “An Environmental Profile and Atlas of Caprivi”, Environment Profiles Project , Directorate of Environmental Affairs, Ministry of Environment and T ourism, Windhoek

• MENJENGUA M. & al. (1995) “An Action Plan for the Development of Rural Development Centres, Monitoring & Evaluat ion Report no 3”, Monitoring & Evaluat ion Unit , Directorate of Planning, MAWRD, Windhoek, July

• MMAPAT SI M.D. & MALEKE J.M. (1996) “Development of an Inst itut ional Support Systems for Sorghum-Mill ing T echnology”, in Drought-Tolerant Crops for So uthern Africa, Proceedings of the SADC/ICRI SAT Regional Sorgh um and Pearl Millet Workshop, 25-29 July 1994, Gaberone, Botswana, Leuschner K. & Manthe C.S. ed., ICRI SAT , Andhra Pradesh, India

• MMIU “Mahangu News” monthly newslet ter of the Mahangu Marketing Intelligence Unit (MMIU), Oshakati, 1999 - 2000

• MMIU / NAB: Minutes of the Mahangu Marketing Intelligence Unit (MMIU) Steering Committee Meetings, Namibian Agronomic Board (NAB), Windhoek, 1996-2000

• MT I (1998) “The Small Business Baseline Survey, Volume 1: The Four Northern Regions”, Ministry of T rade and Industry (MT I), Windhoek, December

• NAB (un dated) “Grain Storage & Handling Manua l”, Namibian Agronomic Board (NAB), Windhoek

• NDT (1998) “Rural Development Centre Ongwediva: Annual Report April 1997 - March 1998”, Namibia Development Trust (NDT)/Directorate of Agricultural Research and T raining - MAWRD, Windhoek, May

• NESANGANO M. & MU ASHEKELE H. (1999) “Strategies for the Management of Appropriate Industrial Production T echnologies: Appropriate T echnology Extension Programme for Selected SME Manufacturers”, consultancy report for MT I and GT Z, Windhoek, (draft) April

• NEW CROPS NEWS (1993), “Pearl Millet : New Feed Grain for Indiana”, vol. 3 no 1, Spring 1993, New Crops home page

• NEWFIU (Namibia Early Warning & Food Information Unit) “Crop Assessment Reports”, Namibia Early Warning & Food Information System, Directorate of Planning, MAWRD, Windhoek, 1990-2000

• NGOLWE A.R. (2000) “Reducin g Post Harvest Losses in Cereal Grains”, in Agri-Info vol. 6 no 2 (14-16), DART , MAWRD, Windhoek, July

• NRC (1998) “Viability of Tractor Services in the Communal Areas”, final consultancy report to NAB & MAWRD, Windhoek, September

• NRC (1997) “T he World Market for Pearl Millet and the Potential for Imports into Namibia”, final consultancy report to NAB & MAWRD, Windhoek, December

• NRI (1997) “Grain Storage in the Northern Communal Areas of Namibia”, Consultancy Report to NAB/DoP-MAWRD, COULT ER J. & HINDMARSH P. (NRI), Chatham, UK, October

• PARMENTIER M. & FOUA-BI K. ed. (1989) “Céréales en régions chaudes: conservat ion et t ransformation” actes du Colloque International de Technologie - Centre universti taire de N’Gaoundéré, Cameroun, 22-26/02/1988, John Libbey Eurotext , Paris

• PINSON G.S. (1979) “A Pedal-Operated Grain Mill”, Rural T echnology Guide no 5, T ropical Products Inst itute, UK

• PIT OIS C. & HELMST ETT ER D. (1996) “T he Set t ing of a Milling Machine at Onamutanda”, T echnical Information Paper No 4, MAWRD/NNRDP, Ongwediva, Namibia, Jan uary

• PRESLAND C. & POMUTI A. (1997) “Towards a Strategy for Sustainable Poverty Reduction in Namibia: Study on the Market for Millet and Millet Products (Mahangu)”, NEPRU/SSD, Windhoek, June

• REPUBLIC OF NAMI BIA (2000) “General Notice No 89 (MAWRD): Grading Standards for Whole and Decort icated Pearl Millet (Mahangu) Grains”, Government Gazet te of the Republic of Namibia, no 2353, Windhoek, 20 June

• RIIC (1997) “Catalogue of Goods an d Services”, 3rd edition, Rural Industry Innovation Centre (RIIC), Kanye, Botswana

• ROHRBACH D.D. (2000) “Commercializat ion of Sorghum and Pearl Millet in Zimbabwe”, paper prepared for the Workshop to Develop a Strategy for the Mult iple Use of Mahangu (Millet) and Sorgh um in Namibia, 27-28 June 2000, Mokuti Lodge,

Namibia • ROHRBACH D.D., LECHNE R W.R., IPINGE S.A. & MONYO E.S. (1999a)

“ Impact from Investments in Crop Breeding: the Case of Okashana 1 in Namibia”, Impact Series no 4, ICRISAT , Patancheru, India

• ROHRBACH D.D., MUPANDA K. & SELEKA T . (1999b) “Commercializat ion of Sorgh um in Botswana: T rends and Prospects”, paper prepared under the project Lessons derived from the rise of small-scale m illing in Botswana funded by IDRC

• RUTEC (1995) “Catalogue - Micro-Enterprise Systems”, Braamfontein, South Africa • SAT News (1999) “Pearl Millet helps the Amazon!”, ICRISAT website, May • SCHADE K., KALILI N. & SIMPSON R. (2000) “A Survey of Subsistence Farmers

in the Ohangwena Region”, research report no 19, NEPRU, Windhoek, January • SE RGEANT A. (1997) “Future Activit ies of the Mahangu Marketing Intelligence

Unit, Interim Report”, prepared for Namibian Agronomic Board, Namibia Resource Consultants (NRC), Windhoek, October

• SIL VE R J. W. (1997) “A Proposal for the Sustainable Millet Production, Processing and Marketing in Namibia”, ‘Millet Development Agency’, Windhoek

• T AMSA (undated) “Catalogue (No 964): Agricultural Milling Machines & Equipment”, Port Elizabeth, South Africa

• T JIPANGANDJARA K. & KIANGI G. (1996) “Appropriate T echnology Study on Food Processing in Rural Namibia ( with special focus on rural women)”, UNIFEM/MAWRD, Windhoek, May

• T JIPANGANDJARA K. & KIANGI G. (1997) “Proposal on Appropriate T echnologies in Food Processing in Rural Namibia, prepared for MAWRD, UNIFEM, NDT , RDC”, (Engineering) Science and T echnology Division, Multidisc iplinary Research Centre (MRC), UNAM, Windhoek, January

• T RANST EC (2000) “Namibia Agricultural Support Services Programme: Appraisal Report”, report for MAWRD and EU Delegat ion in Namibia, draft , August

• T YLER P.S. & BENNETT C.J. (1993) “Grain Market Liberalisat ion in Southern Africa: Opportunit ies for Support to the Small- Scale Sector”, NRI, Chatham, UK

• UNICEF / MAWRD (1999) “Extension Staff Handbook Volume 2: Crop Production”, Windhoek, May

• VAN DEN AKKE R S. & GI GA D. (1992) “Manual on Food Grain Post-Production T echnologies and Pract ices for Small-Holders in Zimbabwe”, FAO/UNDP/Agritex Project ZIM/88/018, Harare, August

• VI GNE P. (2000) “Report on an Evaluat ion of the Activit ies of the Mahangu Marketing Intelligence Unit” presented to MAWRD/NAB, Windhoek, 21 February

• VI GNE P. & al. (1997) “Rural Development Centres: Ongwediva, Ben Hur, Mashare ARDI, T sumis PAK: a Review and Recommendations for the Way Forward”, OXFAM/MAW RD, Windhoek, August

• WANDERS A.A. & HEIJBOER J.P. (1991) “ Introduction and Local Manufacturing of the Votex Ricefan Thresher, experience at ‘Office du Niger’ Arpon Project - Mali”, Directorate General for Internat ional Cooperat ion, T he Netherlands, November

• YARON G. & al. (1992) “Rural Development on the Okavango Region of Namibia”, NISER (UNAM) & Oxfam Canada, Gamsberg Macmillan, Windhoek